Hubert H. Humphrey Building
Room 505A
200 Independence Avenue, SW
Washington, D.C.
DR. COHN: Welcome, everyone. We will get started. Good morning. I want to call this meeting to order. This is the first day of three days of hearings of the Subcommittee on Standards and Security of the National Committee on Vital and Health Statistics. The committee is the main advisory committee to the U.S. Health and Human Services on national health information policy.
I am Simon Cohn, Chairman of the Subcommittee, and the National Director for Health Information Policy for Kaiser Permanente. I want to welcome subcommittee members, staff and others here in person. I do want to mention that normally, we are on the Internet. My understanding is that the connection with the DA who hosts us for these sessions is still being established, and we will let everyone know when we actually are on the Internet.
Obviously there is a lot to cover over the next three days. Today we continue our work regarding e-prescribing standards As all of you know, the Medicare Modernization Act calls on the Secretary to adopt standards for e-prescribing, and the NCVHS has been directed to develop such standards recommendations.
We start the day with presentations and discussions with e-prescribing networks. Thank you for being here. After the break, we are fortunate to have a presentation from Doug Bell from Rand on the recently completed study on e-prescribing. After lunch we have David Bates coming in to talk about patient safety as it relates to e-prescribing. Then after the afternoon break, we have implementers who will show their thoughts on other e-prescribing standards and issues.
Finally, somewhere between five and six, we will finish with the testimony, have open mike, and then have subcommittee discussions. So clearly, we have a full day.
As mentioned, the hearings will continue all day tomorrow, and then on Thursday they will continue in the morning and then adjourn around 1 p.m. I want to thank Jeff Blair, our vice chair, for his leadership in moving forward with the e-prescribing work plan, and Maria Friedman, our lead staff, for her valued and tremendous efforts putting these hearings together.
I want to emphasize to all that this is an open session. Those in attendance are welcome to make brief remarks, as long as their comments are germane to the topics at hand. We will also have time at the end of the afternoon for brief comments during the open microphone session for those in attendance. Finally, when the Internet is up and running, we welcome generally e-mails and other communications in writing from those in attendance or in the Internet.
With that, let's have introductions around the table and then around the room. For those on the National Committee, I would ask, if you have any conflict of interest related to any issues coming before us today, would you please so publicly indicate during your introduction. Maria, maybe you would like to start the introductions.
MS. FRIEDMAN: Thank you, Simon. I am Maria Friedman from the Center for Medicaid and Medicare Services, and lead staff to the subcommittee.
DR. WARREN: I am Judy Warren, a member of the subcommittee, and work at the University of Kansas.
MR. REYNOLDS: I am Harry Reynolds, vice president of Blue Cross Blue Shield of North Carolina and a member of the subcommittee.
MS. TRUDEL: Karen Trudel, Center for Medicaid and Medicare Services, staff to the subcommittee.
MS. PICKETT: Donna Pickett, National Committee on Vital and Health Statistics, Centers for Disease Control and Prevention and staff to the subcommittee.
DR. SCANLON: I am Jim Scanlon, HHS Office of Planning and Evaluation, and I am the executive director of the full committee.
DR. GUINAN: Jack Guinan, CTO, ProxyMed.
MR. BRADLEY: CEO, RxHub.
MR. HUTCHINSON: Kevin Hutchinson, CEO, SureScripts.
DR. WHITTEMORE: Ken Whittemore, Vice President of Professional and Regulatory Affairs for SureScripts.
MS. AMATAYAKUL: Margaret Amatayakul of Margaret A Consulting. I am assisting the subcommittee.
DR. HUFF: Stan Huff with Intermountain Health Care and University of Utah in Salt Lake City, a member of the subcommittee. I don't know that I have any conflicts for the things that are going to be discussed today.
DR. STEINDEL: Steve Steindel, Centers for Disease Control and Prevention, staff to the subcommittee and liaison to the full committee.
MR. BLAIR: Jeff Blair, vice president of the Medical Records Institute, vice chair of the Subcommittee on Standards and Security. I'm not aware of any conflicts of interest related to the topic that you are describing.
MS. SQUIRE: Marietta Square, HHS, CDC and staff to the subcommittee.
MR. MANTOOTH: Mark Mantooth, Health and Human Services Office of General Counsel.
MS. HOWARD: Cynthia Howard, CMS.
MS. GILBERTSON: Lynn Gilbertson, National Council for Prescription Drug Programs.
MS. CANFIELD: Ann Canfield, RX Benefits Coalition.
MS. THOMAS: Ceil Thomas, The Pink Sheet.
MS. PARKER: Wendy Parker, Camarick.
MR. BOSCOWITZ: Roy Boscowitz, National Association of Chain Drugstores.
(Whereupon, the remaining introductions were performed off mike.)
DR. COHN: Thank you all for joining us. Jeff, before we start our session, would you like to make a few introductory comments?
MR. BLAIR: Sure. Thank you, everyone, for being here. I am going to just say a few background remarks.
I am assuming all of you are aware of the fact that this initiative is being driven by the requirements in the Medicare Prescription Drug Improvement and Modernization Act. In that act, it wound up requesting that the NCVHS explore the areas of standards for e-prescribing. The law was especially specifically in terms of the areas of standardization that we should consider and the number of stakeholders, or not the number, but the specific names of the stakeholders that we should listen to. l thought the list was long, there were some additional stakeholders that we felt were important, that we have also added to that list. Many of them are here today and tomorrow for us to hear from.
For those of you who are looking for a little bit of a perspective on how the NCVHS is approaching this task, there is a working document. It is a work plan. It is version eight right now. Most of you -- I don't know if I can say most of you; many of you obviously have gotten copies of that work plan, but if there is anyone who hasn't, please ask either Maria Friedman or myself or Marietta Squire, and we will be more than happy to get you a copy of it. That is a work in progress. It will be changed and updated, but at least it will give you some feeling for the directives and where we are going forward.
Now, in addition to the oral testimony today, the NCVHS has asked all of the testifiers to respond with written testimony as well. There is a lot of important information in the written testimony which, to be honest with you, is too detailed to wind up using that guidance for the testimony as guidance for oral testimony, but there is important information in there, and I think many of you would be interested in that. So again, either Marietta Squire or Maria or myself, let us know and we will get you copies of that.
Other than that, I think you all have seen the agenda, so I am not going to repeat the agenda. I am just going to say welcome, and I am looking forward to hearing from all of our testifiers today.
DR. COHN: I do want to announce that we are now on the Internet. For those on the Internet, we have gone through introductions. This is the Subcommittee on Standards and Security of the National Committee on Vital and Health Statistics, and we will be starting our first session on e-prescribing networks.
With that, Kevin, do you want to lead off?
MR. HUTCHINSON: Good morning. As I mentioned, my name is Kevin Hutchinson. I am the president and chief executive officer of SureScripts. Accompanying me is two of my SureScripts colleagues, Bob Beckley and Ken Whittemore. Speaking for our entire organization, I thank the committee for inviting us here to share our experiences and conclusions, gleaned from an ongoing effort to deploy electronic prescribing nationwide.
SureScripts was created by the National Community Pharmacists Association and the National Association of Chain Drugstores in 2001 to improve the overall prescribing process. Working with the nation's premier health care technology vendors, SureScripts is creating an open, neutral and secure system that is compatible with all major physician and pharmacy software systems.
More than 60 percent of the nation's retail pharmacies have now tested and certified their pharmacy application on the SureScripts network. That number is expected to grow to more than 75 percent of the pharmacies in the U.S. by the end of the summer, 2004.
In the fourth quarter of last year, SureScripts launched a national rollout of electronic prescribing in pharmacies all across the country, and are in various stages of activating those stores for e-prescribing connectivity, now that the majority of the pharmacy software certification and testing is complete.
Since the beginning of 2004, in coordination with the pharmacies, we have activated electronic prescribing in eight states, with more than two dozen states currently scheduled with go live dates before the end of 2004. Additional markets will continue to go live in 2005 and through 2006. We are out of the planning stage and into the execution phase for electronic prescribing connectivity.
As I just mentioned, SureScripts was created to improve the overall prescribing process. This is an important point that I would like to touch on for just a moment. We found that all too often, the popular but narrowly focused term, e-prescribing, has caused confusion and misunderstandings about the true scope of what we hope to accomplish for patients and the health professionals who care for them. The prescribing process does not begin when the physician first touches the prescription pad, nor does the process end when the pharmacist ends the medication to the patient.
Looking at the prescribing process from the standpoint of the physician, one can see there are numerous steps that occur before the creation of the prescription. The patient's chart is pulled and reviewed, the patient is interviewed and examined, a diagnosis is decided upon, and a course of therapy is contemplated and then decided upon. If it is decided that medication therapy is an appropriate choice for the patient, it is at that point that a prescription is created, and also noted in the patient's chart. When it comes time to authorize a refill renewal request for the patient, many of these activities are repeated. All in all, considerable time, expertise and judgment are invested in these activities. We believe there are several points in the process that might be improved beyond the simple act of generating a prescription.
At the pharmacy end, much more is involved than generating a prescription medication than simply placing tablets or capsules in a vial. Electronic pharmacy patient records, which include allergies, existing medical conditions must first be created. Prescription insurance information must also be inputted and updated periodically. Upon the receipt of a prescription for a patient, prescription infrastructure is inputted in the pharmacy computer, which immediately performs a drug utilization review against the medications listed in the pharmacy record. Once the pharmacist has reviewed any DUR issues flagged by the pharmacy system, the prescription is billed to the insurer. During the billing to the insurer, an additional DUR check is performed by the insurer, and the resultant payor issues, either financial, claim or clinically related, are resolved by the pharmacist and the prescription is prepared for dispensing. The process is then concluded when the prescription is dispensed to the patient, and the patient is counselled on its use by the pharmacist.
My point in going into all of this detail is to emphasize to the subcommittee that our goal must be to improve the overall prescribing process. To focus too narrowly on the act of generating a prescription and transmitting it to a pharmacy would ignore many opportunities to enhance the level of safety and quality of health care delivery to Medicare patients. It would be our pleasure to assist the subcommittee in working toward this goal.
I would now like to address the specific questions about standards for prescribing that were proposed to us by the subcommittee.
You first asked what are the best standards for code sets to meet the requirements of the MMA, which do we use, what are their strengths and weaknesses, and is their nationwide adoption necessary.
MR. BECKLEY: In implementing our electronic prescribing connectivity system, we selected the nationally recognized NDPDP script standard to serve as the foundation of our transaction engine software. The NCPDP script standard was developed to facilitate the electronic bidirectional transmission of prescription information between prescribers and pharmacies.
It is our experience that the use of the script standard improves patient safety, quality of care and efficiency, without presenting any undue administrative burden on prescribers and pharmacies. We believe that the NCPDP script standard is the best standard to meet the e-prescribing needs of Medicare patients and physicians and pharmacists who serve them.
The script standard was developed through a consensus process among community pharmacies, community pharmacy organizations, pharmacy software vendors, database providers and other stakeholders. Further, the standard addresses the electronic transmission of new prescriptions, prescription leaflet requests, prescription fill status notifications and cancellation notifications, the nuts and bolts of e-prescribing.
Further enhancements could address other possibilities that may include patient eligibility, compliance, lab values, diagnosis, disease management protocols, patient drug therapy profiles and/or prescription transfers.
The strengths of the NCPDP script standard are that it is a national standard that addresses the vast majority of the core functionality required in the MMA. It currently facilitates the bidirectional transmissions of prescription information between prescribers, dispensing pharmacies and pharmacists, and holds the potential to allow for the transmittal of information eligibility and benefits and medication history.
A weakness of the script standard is that it is not widely used in the inpatient and long term care settings. However, to address the weakness, NCPDP and HL-7 are initiating a collaborative effort to harmonize the two standards with respect to e-prescribing.
In terms of the best code sets, we found that the NCPDP provider identification number, which many pharmacies adopted several decades ago and is used universally in the processing and payment of prescription drug claims works extremely well as the pharmacy identifier code set for e-prescribing. It was formerly known as the NABP number.
The strength of the NCPDP provider identifier is that it is used nationally to identify the participating community pharmacy. This unique identifier is used by boards of pharmacy, payors and telecommunications switches to precisely identify pharmacies for a variety of purposes. This identifier has been used effectively by the stakeholders for several decades and has been refined substantially over that time. We do not believe that it has any weaknesses.
MR. HUTCHINSON: With respect to packaged drug code sets, we believe that NCVHS should consider the NDC number as the best standard. The NDC number has also been used for several decades, both for prescription drug claims processing and to facilitate internal data processing needs in community pharmacy computers. SureScripts transmits the NCD number when submitted by either prescriber or pharmacy software.
The strength of the NCD number is its universal acceptance as a packaged drug code set and its use by payors, including CMS, pharmacies, manufacturers and database companies. The primary weakness of the NDC number is that it does not specify component ingredients in a drug product, and thus is not useful as a drug ingredient code set. Even when an NDC number is used in combination with proprietary drug ingredient databases to meet this need, there are problems, because there are no standardized references between such databases.
All three of these standard code sets, the NCPDP script standard, the NCPDP provider identifier for pharmacies, and the NCD number as the packaged drug code set, are recognized and adopted nationwide.
The subcommittee also asked if there if there are any e-prescribing standard or code set gaps and what barriers might prevent their development and adoption.
The first e-prescribing code set that SureScripts encountered was that of a prescriber identifier code set. Code sets that have been suggested as useful for this purpose include the prescriber's DEA number, NCPDP's HC ID and the MPI, all of which have significant deficiencies.
Because there is not a patient identifier code set that uniquely and adequately identifies prescribers on a national basis, SureScripts was compelled to create its own SureScripts prescriber ID or SPI for internal use. The SPI consists of an SPI route and a location ID. Because a physician can prescribe for more than one location, the location ID is used to distinguish between multiple locations for the same provider.
This is important, because the proper mounting of refill renewal requests and prescription fill messages is not possible without location information. It is in this way that SureScripts has been able to overcome the location deficiency of the NPI.
Another e-prescribing code set gap is a nationally accepted non-proprietary prescription direction, or SIG database. Because of this, SureScripts must now transmit prescription directions in free text format, which compromises a portion of the efficiencies that pharmacy providers had hoped to achieve when implementing e-prescribing.
The lack of a standard patient identifier is also a gap in code sets when it comes to e-prescribing. It is unlikely that either medication or medical histories will be implementable without a nationally accepted standard patient identifier code set. Current methods of trying to identify a patient are done through complex algorithms which add to deployment costs and vary in accuracy. Either a standard nationwide patient identification system needs to be established or a master index needs to be maintained.
As to drug ingredient code sets, there needs to be a cross reference between proprietary drug ingredient databases, as the NCD number was not created to list individual ingredients.
Finally, the subcommittee asked, what incentives or other suggestions should the government consider to accelerate the development and/or adoption of e-prescribing standards or code sets. As the subcommittee is aware, the MMA recognizes the financial challenges that e-prescribing will likely represent for prescribers by providing a grant program that will provide a 50 percent financial match toward the purchase, lease or installation of computer software and hardware, upgrades or other improvements to existing software, and providing education and training to prescriber staff on the use of e-prescribing technologies.
The widespread implementation and use of the developed e-prescribing standards and code sets could be further accelerated by expanding the grant program to pharmacies, which are supporting a disproportionate share of the overall e-prescribing infrastructure and transaction costs.
Further, we recommend that the 2006 pilots be used to assure that the new standards mentioned in the MMA do not impose an undue administrative burden on health care professionals involved in e-prescribing.
It is imperative that these standard code sets be proven to work in the private sector prior to requiring their national use. Again, to further accelerate the adoption and use of e-prescribing standards and code sets, SureScripts recommends that the financial provisions of the MMA be expanded to include pharmacies.
MR. BECKLEY: In closing, I bring up what we believe to be one of the most critical aspects of insuring widespread adoption and integrity in the electronic prescribing process. As I mentioned in the opening, SureScripts was established in the pharmacy industry to insure an open and neutral secure system for the widespread adoption of electronic prescribing. In fulfilling that mission, SureScripts has created certification rules or standards that preclude any technology partner connected to the SureScripts network from displaying commercial messages with the intent to influence a physician's decision in medication therapy or with the intent to influence or alter a patient's choice of pharmacy, be that a chain pharmacy, an independent pharmacy, a supermarket pharmacy, mass merchandiser pharmacy or mail order pharmacy. However, of course, we believe in and our rules support the use of, formulary management at both the point of care at the pharmacy and the point of care with the physicians.
This approach is not dissimilar to the approach of the banking industry as taken with the deployment of electronic banking across the country, for the banking industry has established standards and rules to be used when interacting with banks electronically. Whether you are paying bills online, electronically depositing salary checks or withdrawing cash from an ATM, banks do not compete to win your business or influence how much cash you take out, unless of course you are taking out more than you actually have, which they will kindly give you a polite message.
The point is that MMA does address the point that electronic prescribing should not in fact allow messages unless, and I quote, it relates to the appropriate prescribing of drugs, including quality assessment measures and systems referred to in Subsection C1B. I believe this to be the committee's largest challenge in defining what is an appropriate formulary or clinical alert message in what is disguised as a commercial message.
In the past year and a half since I have been at SureScripts, I have personally witnessed some very interesting business models by companies, where the belief that large sums of money can be made is a message appropriately or inappropriately, physicians with the intent of switching medication therapies or a patient's choice of pharmacy.
I'm sure the banking industry realized early on the value of the consumer eyeballs on their ATMs, online checking and personal banking software connected to their banking networks, and elected to err on the side of protecting the integrity of the electronic process and maintain consumer confidence in the system. A high level of automation existed in the banking industry prior to the deployment of electronic banking. The challenge was to get consumers and companies to connect to them electronically to improve the efficiency and the security of the banking process. Processes were created to leverage the existing technology by extending the process out to the consumer.
This is very similar to the opportunity before us today in the health industry. A high level of automation exists in pharmacies today. Close to 100 percent of pharmacies have automated their clinical and financial claims process. As an industry, we are exploring the possibilities to leverage not only the existing technology, but also build upon the relationship that exists today between a patient's pharmacist and a patient's physician, to enable the two health professionals involved in the process to make a safer, more efficient and potentially higher quality medication decision for us all as patients.
We at SureScripts thank the subcommittee for the opportunity to share our experiences and findings with respect to standards currently used in the United States to facilitate the adoption of bidirectional electronic prescribing communications between prescribers and pharmacies. We encourage NCVHS to support current e-prescribing efforts that have evolved using these standards, and build on ones that need further clarification and/or description.
Thank you for your kind attention.
DR. COHN: Thank you very much. Jim Bradley, I believe you are next.
MR. BRADLEY: Jack has agreed to run the slide show here for me this morning.
First to the subcommittee, thank you very much for allowing RxHub to be allowed to participate. Before I start my remarks, I think it is important that one's perspective be put into a context. I was employee number one at RxHub, but that was after a number of years, in which I would characterize most of my backgrounds being one of a payor.
In addition to that, I am old enough here over the many people in the room, and so I am old enough to remember the days when there was no drug benefit in a lot of the commercial insurance programs, and I got to watch what happened as we started the provider drug benefit, and looked at such issues as adverse selection, one of open access to a benefit that didn't exist before. I went from there into a belief that we were going to have to actively engage the physician and patient as we ran the health care system moving forward, because without that, we had a lot of inefficiencies.
I also got to watch what happened when first there was a catastrophic drug benefit that incented automation. Then there was an incenting by HHS for an electronic claims submission for Medicare claims. I got to see what happened to us in the commercial sector, meaning game over. It accelerated the process in such a way that we commercially were able to avail ourselves of high degrees of electronic submission for commercial claims.
As we started the job at RxHub, we felt there would be several bellwether events, if we were to achieve the dream of having an informed patient and physician making rational economic and clinical decisions at the point of care. One of the bellwether events was that the physicians had to adopt technology. That was going to require a healthy technology sector.
At the time, and this was 2001, you will recall that we had a dot-com bubble that had burst, a lot of early adopters who had been burned, and a lot of cynicism. So we felt that until some of these firms re-entered the marketplace, we had no hope for automation.
The second bellwether event, we felt, was going to be some form of drug benefit passing for our senior citizens. The reason for that is that in most physician practices, as you go in and try to change the way they work, what you find is, you can only represent about half the prescription buy-in in that clinic. So most tools that need to get adopted wouldn't represent enough of the physicians' business to make that kind of change.
So last year with the passage of the Medicare benefit, we felt, now we are in a new world. It is uncharted. It is an unprecedented benefit.
As I go through the slides now, I want you to remember that perspective. Our perspective at RxHub has been as much around the affordability of a benefit, as much as just the electronic automation of it, and making sure that the information was available in the physician office to allow for that kind of informed decision making.
I won't repeat a lot of what Kevin said. I think a lot of the early work with e-prescribing took a very simplistic approach. The simplistic approach was, how do you get a prescription from a physician office to the retail pharmacy where it is dispensed. The problem with that approach is, if there is anything flawed with today's process, it just automates the flaws. So we felt that it was very important to dissect the work flow process in a physician office, and we broke the process down into these four areas. There is a part we call prescribing. That is the decision support effort on the front end of a prescription that helps the physician and patient come to a conclusion as to what is appropriate therapy.
The second is the actual writing of the order. The writing of that order needs to be done in such a way that it takes advantage of that information that is on the front end. There is the delivery of that to the patient, then finally, somehow it gets to where the prescription is going to be dispensed.
Now, just to be clear and go on the record about this, we fundamentally believe at RxHub in patient choice. The patient should be able to choose the drug therapy, they should be able to choose where that gets dispensed. However, in a rational system that provides this level of coverage, those choices are going to have certain costs and tradeoffs. We believe in an informed choice to be made at that point of prescription and fulfillment.
So as we go through the definitions of e-prescribing, we say that it has a number of things. Access to clinical decision support. We believe that sharing of the patient history is important to the clinical decision making process. We think that we need to increase the practice efficiency of physicians in both the inpatient and outpatient settings, and also in providing an efficient business framework. So this is really an information problem more than anything else that we are trying to solve here.
In the Medicare legislation, there are a number of provisions that I'm sure everyone in this room is familiar with, but in case you are not, it starts with a process of determining eligibility for the drug benefit. So the first piece of information that is required is something about eligibility, what that benefit looks like, including formulary and preferred drug and tiered benefit structures and so forth need to be understood.
There needs to be clinical information on the drug being prescribed, and information about the availability of lower cost therapeutically appropriate alternatives, meaning that the data has to be rich enough to provide that patient and physician with a rational set of choices.
Absent this information, this benefit has the opportunity to bankrupt Medicare. In other words, I watched what it did in the commercial sector. Without some form of structured approach to the benefit and the decision making, we have a financially unviable benefit here.
So what do we need? Part of the problem we have today is that we have a general lacking of an infrastructure to provide this kind of information. I think that we start with the first level problem being patient identification. Again, by matter of perspective, some of you and I worked together back in the early '90s in WIDI after Secretary Solomon initiated that work effort with a number of industry colleagues; we were trying to address the administration simplification process.
Back in 1992, we talked about the need for a patient identifier that could easily and uniquely get you to the right individual for the storage of both personal and clinical and other kinds of information. I think it in 2004, we are still not there. So if we wait another 12 years without being able to identify the patient, we are going to have a real problem. We are fairly cynical about the ability to get this done in any kind of near term time frame, particularly as this benefit rolls out.
The second thing is, I think you have to talk about an enrolled population. What we are talking about when we say enrolling a population, we just passed a drug benefit that should allow us to enroll seniors now into organized programs, and that enrollment process is the way one organizes for this patient identification process.
Once you have patients in an enrolled database, you can make connections to everywhere else. Absent that enrollment process and that directory process that gets you to the patient, most efforts to consolidate information, interoperability and other things, the systems break out. If you can't identify the patient and then figure out where the data is, you cannot organize an infrastructure that is feasible.
Here is how we approached at RxHub the patient identification process. Remember, as we started RxHub in 2001, at the point in time we did not have a drug benefit. I was not personally given the latitude to wait for a unique patient identifier for all citizens. I had to get going.
So how do we approach the problem? In the three PPMs that founded RxHub, we had approximately 150 million lives. So what we did is, we ran those 150 million lives through statistical algorithms to identify those data elements which could uniquely get us to the patient and have no false positives. In other words, if we found Jeff Blair's evil twin and delivered a medication history, we had a problem. We had to find the right Jeff Blair.
At times in the world of HIPAA, you might like to present the potpourri of alternative Jeff Blairs to the physician and let them select one. You cannot do that. So there are times when we say we cannot find the patient when we really have them in our database, but that is just part of the problem here that we have to put up with.
We know after millions of transactions through this MPI, we know of no false positives, meaning we have asked physicians who are looking on the other end to verify that we have the right patient, and we know of no instances of a false positive.
So 150 million lives, five elements that you can get off their driver's license, no false positives, and we can identify those people in less than a quarter of a second. We think it is no issue for a Medicare benefit to be built around a commonsense approach like this one to identify seniors who are in this program.
Another thing about this MPI is, it immediately tells us about coordination of benefit opportunities. For those of you who aren't from commercial insurance backgrounds, COB is nearly impossible for drugs because of the high volumes and the low returns of the COB effort. This allows us to immediately identify those multiple coverages and adjudicate them.
The final point. We do not store patient information. We do not store anything other than these five data elements and a unique identifier at RxHub. So we don't have to worry about competing companies sharing their customer master files. That is a big hurdle to get over. We just store enough elements to identify the patient, and then point at where the drug coverage is.
For eligibility, coming at that problem, we brought a payor's perspective to it. WIDI a long time ago recommended the ANSI standard set for eligibility. Therefore, most systems that operate in physicians' offices today understand the ANSI transaction set happens to be the X-12 70 and 71 transaction. That is the one that RxHub suggests for the eligibility process in the physician's office. It is simpler to adopt a variant of that for drug benefit eligibility than to invent a new transaction.
It also is the one suggested by HIPAA for eligibility. There are other transactions, some of which come from NCPDP. A number of NCPDP transactions we will talk about here, but for this one, it is not widely used for eligibility by physicians, and we think it is harder to adopt than ANSI X-12. So it is nothing against the NCPDP transaction; it is primarily tailored to pharmacies. Most physicians' offices do not have the capability.
Medication history. This was something that has emerged over the last year. Actually, our work with medication history started before RxHub existed. At a startup I did, we worked with Medco at the highest volume prescribing site, and found that after automating the physician office, the number one thing that the physicians found helpful to their practice was the medication history that could be delivered.
This medication history when it comes out of a payor database basically has all of the prescriptions that are being paid for, regardless of the site of dispensing, regardless of who prescribed them. At RxHub today, these transactions are alive and well. We can deliver a medication history always less than ten seconds, most of the time less than five seconds.
The medication history, we believe, is number one sticking point for physicians. In other words, you can automate a physician office to improve their efficiency. You also need to improve the quality of care that is being delivered. Physicians resonate to this history, and we think it is important.
No standard for medication history existed, so RxHub developed one. We are in the process. We used as much of the NCPDP transaction set as we could to deliver this, and we are in the process of taking that standard through the NCPDP process. We do believe that the nationwide adoption of medication history transactions is required for the patient safety objectives that come out of this bill as well as other bills and other commercial efforts such as Leapfrog and so forth.
In terms of how you get to the medication history, I think after we developed the transaction and put it in production, a lot of people said this is a great idea. We have got medication histories too, and we will all compete with each other to deliver medication history.
The problem with medication history though and doing it feasibly, you have to deliver it in seconds, not minutes, not hours and so forth, and you have to very quickly and reliably find the information.
What I would suggest is that once the senior population is enrolled and alongside the commercial population, the payor database that has the paid information about prescriptions that are delivered is the most complete to start the medication history, meaning, in that case you have all prescriptions that have been paid for. The ones you would not have are those that are paid for in cash. You would not have those that fall oftentimes under a copay screen. So if the drug is less expensive than the copay, oftentimes the pharmacist will charge the patient the price of the drug and not enter a claim. You would not have those. You would not have OTCs.
To enhance that medication history, there is an opportunity to cooperate between the payor database and the pharmacy database to add those additional drugs to the medication history. But remember, we have got to do it in ten seconds or less. The good news is, most payor databases would have not only the information about what has been paid, but they also would know who the prescribing physician is, they would know who the dispensing pharmacy is, and if there was an infrastructure available to seek that medication history from the pharmacy, it could significantly enhance what we are talking about here.
So we believe in a more or less hierarchical approach. You start with a patient ID n 2.5 seconds, you add to that with a medication history in less than three or four seconds, and then you initiate and extend the transaction to the retail pharmacy to augment the information.
We have been putting these medication histories into clinical settings now for about a year. We have gotten nothing less than rave reviews. It is information that clinicians have never had before.
Benefit coverage. There, we are talking about the kinds of things that one would expect in a benefit transaction, including formulary, preferred drug status, availability of generics and so forth. The information required for this, we went through a public work group process and developed a standard for a transmission of formulary. The formulary information has to be available in two ways. Some technology vendors like to download the formularies so they are present on their systems. Others want to actually have a real-time transaction to verify formulary status. Both of those standards have to be available. We are in the process of taking that through the standards processes as well.
Is national wide adoption necessary? Absolutely. The benefit coverage has to be known at the point of care before these decisions are made.
For prescriptions, the prescription process is complex. It is not simply a renewal request and a change request, but there has to be a new prescription transaction, and there has to be all of the supporting information that goes around that.
There have been no standards for a number of aspects for new prescriptions. RxHub developed that again through our work group process. We intend to take that through the NCPDP process as well.
In terms of physician adoption, I think there are a number of things to be accomplished. To start with, at the passage of the Medicare bill, we saw a notable uptick in activity among the technology vendors. So a number of us were sitting on the sidelines, wondering whether to get into the physician market again. That all changed with the passage of the bill.
One of the laments that comes out of the technology sector is, there is no adequate business model for physicians to adopt these products. So what they would say to you is, physicians are cynical. They say, we don't get paid any more whether we adopt tools and enhance patient care and save money or not, so therefore, what is our ROI for adoption of the tools.
So on the commercial side, we are exploring incentives from payors that would provide incremental reimbursement for office visits where these tools were used, and potentially, lower reimbursements where there was a reluctance to chance and adopt the tools. So we believe that such an incentive is important.
During the legislative development process, there was talk about, in the Medicare benefits itself, should there be a mandate which would require electronic submission or the drug wouldn't be paid for. The problem with that approach would be that the person who would be not adopting would be the physician. The drug not being paid for would be for the senior. That is not an appropriate incentive. So we think that the incentive needs to be built around the reimbursement to the physician, and it ought to be a positive incentive that provides an ROI.
So in summary, our industry does have to have standards. We believe that this is a very unique time. We have no doubt that this effort will succeed. As Medicare enables a benefit that covers over half the prescriptions in the United States, standards will evolve.
If the committee decides upon a standard that is different than one that RxHub uses, we will very quickly adopt your standard. So this is not a fight about anything. We are just trying to present what we think are the business challenges around the standards process.
We think the patient ID problem is not going to get solved any time soon. Unless you have some way around it, all this stuff doesn't work, period. So we have talked to everyone from Mark McClellan and the various people who are drafting the legislation and so forth about our MPI process, and we think that that has to be a part of the solution here. So we would suggest that there is a lot that we can offer there.
The third point. We must create an efficient market through which prescription drugs are purchased and delivered to consumers. By efficient market, we would suggest that today with the physician and the patient not having any idea most of the time what the benefit in place is, how can they make a rational clinical and economic decision. While we understand the opinion that suggests the patient can have anything they want and the physician can prescribe anything they want, that kind of benefit is unaffordable. Therefore, we must be able to provide a more rational economic framework to deliver these drugs.
Second to last point. Enough of the industry lack of cooperation. All of these efforts should not be used to compete or for self interest. Our health care system is broken. It is time we start working together collaboratively and build in the kinds of capabilities that we all need, and now it is time to get on with the job.
To the subcommittee, thank you very much for allowing us to come today. We appreciate it. Thank you.
DR. COHN: Thank you very much. Jack, I think you are the final presenter.
DR. GUINAN: Yes. Good morning, and thank you for having myself and ProxyMed present to the subcommittee this morning.
I as well thought I would start with providing a little background about myself and our company as a context for the remarks that I am going to make this morning. I am the chief technology officer at ProxyMed. ProxyMed is one of the primary health care EDI networks in the country serving physicians, so it is my job really to insure that we move about a billion transactions a year through our network. There are some very nuts and bolts kinds of points that I would like to make to the subcommittee today.
We do serve about 150,000 physicians, 800 payors, insurance companies, 30,000 pharmacies, 2,000 labs. We work with about 400 practice management systems. We are a combination of many acquisitions. Our experience goes back more than 15 years or so with health care EDI. We have been involved directly with the prescribing for over ten years. Myself personally, when I started with ProxyMed about ten years ago, this was the basis of the company.
The point of all this that to accomplish many of the goals that have been talked about already this morning and have been talked about in the various e-prescribing initiatives, there needs to be a solid foundation for the EDI movement of information between the various parties, and there are some very important points and goals that need to be accomplished to make that happen.
For the purpose of my discussion this morning, I am going to narrow down what I am going to speak about with e-prescribing. I'll categorize them into three pieces first, the efficient online routing of new prescriptions and refill authorizations between physicians and pharmacies.
Although it has been talked about already this morning that this really is a very basic function and not the complete breadth of what we are talking about with e-prescribing, up until now it has formed the majority of the work that has gone on with the movement of e-prescribing information, and will form a very important basis for the work in the future. This very simple goal does provide quite a bit of cost savings through efficiency for all the parties involved. So part of the goals of the act was to provide cost savings.
We have found that ProxyMed as well as many of the participants in the industry, through documented studies, that the automation of moving new prescriptions and refill authorizations between physicians and pharmacies can substantially reduce the cost of this function for both the physician office and the pharmacy. At the basis of this is replacing phone calls with an online transaction. It does as well have a patient safety effect through the legibility and accuracy of the prescription.
So although many times in these discussions this is seen as maybe too narrow of a goal or a small goal, this is a very important part of what needs to be accomplished. Yet there are some holes to even accomplish this, which I will speak about in a minute.
The next part of the way we look at e-prescribing, providing patient medication history, patient eligibility information and other benefit information is as a distinct part of e-prescribing with its own challenges, as Mr. Bradley just spoke about. Then there is another part, which is to influence which drugs are prescribed, both in a good and a bad way, but to facilitate many of the other goals specified in the act.
What I am going to speak to in the short period of time that I have is about the first point, which is about the routing of the transactions between the various participants.
Again, in routing the transactions, approximately a million a day, there are a couple of basic elements that need to be present. The routing of these transactions is the same, whether we are talking about claims, ERA, prescriptions, lab results.
We have gone through quite a bit of struggle with the recent HIPAA transformation in the claims industry. Many of the struggles we have revolve around a very simple concept: Uniquely identifying the sender and the receiver of the transaction sets.
For e-prescribing, from the viewpoint of ProxyMed and the folks that we work with in the marketplace, one, it has already been mentioned that the pharmacy facility IDs are well established. So to get that out of the way, and as Kevin Hutchinson mentioned before and Bob Beckley, the NCPDP number has been very well established as the pharmacy facility ID, whether that is the sender ID or the receiver in the transaction set. So we feel very comfortable with using the NCPDP number in this very important role.
However, on the physician side of the equation for e-prescribing, this we see as a very large problem at the moment. Physician office IDs are not standardized at all. As was already mentioned, the prescriber IDs themselves are also not standardized.
The way the ProxyMed views the routing of e-prescribing messages is that these are not routed by prescriber. In fact, the routing of these messages are done by the physician's office. For example, in the current system today, you go to the physician. One of my children goes to the pediatrician, they get a medication. It gets billed at Dr. McKee's office. They have many doctors there. They might see Dr. A one day, Dr. B the next time when they go, but all of these prescriptions come from Dr. McKee's practice.
They get sent to the pharmacy on the corner. It gets reported in the system as prescribed by whichever doctor, Dr A or B, but very importantly in the system, the phone number of the practice is reported. When I go back for a refill of my prescription for my child, when they want the authority, they are not contacting the physician who wrote the prescription, but instead they are calling the office where the prescription originated. Many times, the prescriber that wrote the prescription is not there that day. In fact, the physician might not work there anymore. But the prescription didn't leave the facility. My child is treated at that physician office. In fact, the routing is done by the phone number of the physician's office and the pharmacy system. This is the way it primarily has happened up until now.
So the point that I am making is that the routing of e-prescribing transactions should be done by a physician office or a health care facility ID, not the prescriber ID. There are different issues with the prescriber ID, and those do also need to be unique, and there are challenges there. But the routing of these transactions need to be accomplished by facility. This has not even been mentioned in pretty much any of the documents. It all seems to focus around the prescriber ID and relating prescribers to different locations, with the concept of trying to have messages follow physicians as they move from location to location. This just hasn't been our experience.
In equating this to the claims EDI practices, this is very much the claims work. We see no difference. We have been doing e-prescribing for ten years, claims for 15 years. Claims are submitted from a submitter ID from a physician office, or if there is physician information attached, it is very important to know the billing physician, but it is not the way that the transactions are routed. This is done through submitter IDs. This is in fact a huge issue we have on the claims side of the business with proprietary submitter IDs across health plans.
So today, the way this is being accomplished, it was mentioned before by Mr. Hutchinson of SureScripts, there are a number of proprietary schemes that are being used. ProxyMed has adopted its own set of proprietary health care facility IDs and physician IDs and the relationship between the both, and then with the pharmacy partners that we work with on the other side of the network from the physician offices, we exchange these proprietary IDs.
When there was only one network out there or a few, and now there are other growing networks, there is already an issue with connectivity between these. For instance, with the SureScripts network, using their SBI number, ProxyMed using their own proprietary facility ID, it would be quite a burden to have the pharmacy system vendors and pharmacies themselves try to keep track of all the various cross references that would identify which is Dr. Jack Guinan's health care facility to be able to route a transaction.
So we see this as one of the big challenges to accomplishing a national role out of e-prescribing, and we feel that more attention needs to be focused on the concept of a health care facility ID, even as it pertains to the MPI. You can establish a unique physician ID, and then say we will relate it to the many physician locations where he works, but if you don't have a way to uniquely identify the location you are referring to as the physicians move, as different physicians are related to the same practice, you will quickly lose track of where these locations are.
Again, a point already made. Physicians come and go from individual practices, but patients remain. What we have found in moving new prescription and refill authorizations back and forth between physician offices and pharmacies is that it is all about where the chart is, where the information about the patient is at the practice. When a pharmacy needs a refill authorization or a renewal authorization, that authorization request needs to go where the information is about the patient. Today, until electronic medical records are pervasive and it is very easy for a physician to get on his cell phone and look up any of his patients, that happens to be in the office where that patient is seen routinely, where his chart is. So that is where the requests need to be routed back to the facilities where the patient was seen.
So at the moment, this is to us a very clear fact of life, that the routing needs to be by facility for these reasons, and there needs to be more efforts in this area to try to figure out if it is through the MPI initiative, okay, but we feel that needs to be brought in slightly to the standardized facilities.
Another very basic part of moving transactions back and forth is about having obviously standardized code sets. You can be able to route information from one place to another, but if the content of those messages isn't clear to the receiver, then the message does no good.
A few of them I thought I would point out this morning. Again, I am pointing out the challenges as opposed to all of the good things in this short period of time, because I feel that is where this time is most useful.
As far as the prescriber code set, this has been discussed already and touched upon. I have already mentioned that the prescriber ID is not for routing of messages, but for completely identifying who is it that authorized the transaction. Currently we use the DEA number. The common practice in the industry for those of us who participate right now from the retail chain pharmacies is to use the DEA number. It was the only thing that was available.
Now, this is frowned upon by the DEA. This was not the intended purpose. In fact, there are regulations that say this is not supposed to be used for this reason. But the pharmacy systems themselves were written to identify physicians by DEA number, and that is the way the majority still are, so it is going to be some period of time as those evolve and get switched out where we can use a different identifier.
The MPI is certainly a viable alternative. It is not yet available, so there is a gap right now. So for those of us in the industry, we are filling that gap with our own proprietary scheme. Very important. If we cannot quickly arrive at a standardized code set -- and maybe it doesn't have to be one. There could be a few, but if there is not a standardized code set for identifying physicians, it will be very hard to roll out this type of EDI service nationally. We can do it in a limited fashion, maybe even get up to ten or 20 percent of the transactions being conducted. But as you approach dealing with hundreds of millions and a billion transactions a year, it is just imperative that these very simple concepts are nailed down.
The drug code sets were already touched upon earlier by SureScripts. At the moment, for e-prescribing -- and when I say e-prescribing, for the purpose of a physician to communicate to a pharmacy a new prescription, they need to identify which drug this is. At the moment, NDC codes -- in fact, in HIPAA the NDC code for a pharmacy transaction was adopted as the identifier for drugs. So the common practice in pharmacy systems today is to use NDC codes.
This is much more than was used for the ordering of drugs. It is really not viable for this purpose for e-prescribing at this time. It is at the wrong granular level. The generics, there are manufacturer specific components to the IDs, so if there are 100 generics available from 100 different manufacturers for the same drug, physicians just do not care about this. They want to see in their drug pick list one drug for that generic. When it goes to the pharmacy, the pharmacy can figure out how they want to purchase it, package size, from the manufacturer.
So there are many problems with using the NCD. The primary one is that it is at the wrong level of specificity for the drug, and creates issues in drug selection lists at the physician side.
At the moment, the way that we fill this gap in the industry is to use free text. This is fine. It is very similar to the manual process, but it does not allow us to accomplish some of the more advanced goals set forth in many of the things just spoken about. If you cannot identify specifically the drug programmatically, you cannot have programs help you assist in decision support and other types of even retrospective review of drugs. So at the moment, in the NCPDP script standard, although there is field of certain qualifiers that allow you to use NDC and other codes for drugs, the primary practice as been to use free text In fact, we have expanded the fields for the drug name, because it includes both the strength and the unit of measure many times in the free text that is sent to the pharmacy.
So there is an absolute need to resolve the drug code set challenge. This is going to be deployed on a national basis, nationally meaning not just geographically, but if we are going to attempt to get the vast majority of prescriptions written to be codified and sent electronically, this is an area that still is not -- there have been discussions for years going on with this, and there are many organizations that deal with publishing drug code sets that have been working together on this. Yet still today, there is not an eyeball alternative.
Rx Norm was in the materials that were presented to us to comment on. This is certainly something that is going in the right direction. This could serve the purpose. But this is a hole that needs to get filled quickly if we are going to really have a broad rollout.
The SIG codes, the directions for use, another area very complicated to solve, but yet something that needs to be done. Today there are no standardized SIG codes that can be used for this purpose. The common practice is to use free text. So the drift of this being that the current practice is, although a well formatted message and standardized, so it is very easy for us to make connections today using the NCPDP scripts format and to get messages back and forth, but the content of those messages for some very important fields are either proprietary numbers or are free text, and it makes accomplishing some of the advanced goals very difficult.
There are conversations going on about codifying the SIGs and coming up with standard SIGs. This is not something at ProxyMed. This primarily involves the moving from transactions from point A to point B. Just pointing out that this is one of the challenges and one of the code sets that needs to be focused on and established for a broad adoption as well as, if you are going to retrospectively go back and take a look at prescriptions or even prospectively look at what it is you should be prescribing, you cannot have programs help you do this if there are free texts in these fields.
Moving on to something which is a little bit happier subject, message formats. If we are moving transactions from point A to point B, we have to worry about who is sending them, the contents and the message format. ProxyMed has been working with NCPDP since the inception -- before script, but since the inception of script for e-prescribing. It is a very well established format. It serves all of the basic purposes for moving e-prescribing transactions from point A to point B. It has been widely adopted for new prescriptions and refill and renewal authorizations.
I would say across the board, I can't think of a participant in the industry right now that is using something other than the script standard for the message format of these type messages. Physician office software vendors, the hand-held vendors have been able to adopt this. It really is no burden on these folks to use this standard. It is a very strict standard. It follows a very basic set of rules. The pharmacy software vendors, likewise.
It has not been our experience ever in dealing with connecting a partner up for a network that they said, the NCPDP script standard is too difficult to deploy, isn't there something else to use. This has just not been an issue, and since this has been going on now for seven years, I take that to mean that it is easy to deploy, and it is not a burden whatsoever. In fact, it is well developed and easy to deploy.
It is mature. Although we are having these hearings now and we have been having hearings like this for awhile, and we are talking about the future, in practice it is a mature standard. We have been working on this for over seven years.
The process by which this standard has been developed and continues to be maintained is a very good one. NCPDP is a very broad organization. There are participants from all aspects of the marketplace that actively participate in work group 11, the script work group, and spend at least four times a year at the quarterly meetings. In between some work groups there is a constant dialogue that has been going on for this many years, and will continue. So in our opinion, to start again and to adopt something else for this purpose, for the movement of the messages, that would be quite a burden.
If we as a network provider had to go back to the software vendors we work with from both sides of the equation and say, well, we are changing the rules of the game, the standard is changing and you have to recode, -- we have HIPAA, which has been for the purpose of the network quite a challenge, and we are struggling just to keep going with the daily business of getting physicians paid their claims, because we switched the basic format of the message. You have hundreds of software vendors having to modify their software, a very risky proposition.
So we would suggest that switching from the NCP to script standard as the basis of the e-prescribing message format would introduce this type of risk into the process again, and would be very negative to the folks that have spent this amount of time getting to where we are in the industry today.
The organization itself, NCPDP, as well as the message format itself, is a very good starting point. It has already been spoken about that it does not cover all of the bases. The format today does not allow the transaction types that are envisioned. The transaction types that are there do not include all of the data elements. There are code sets missing. Given that and given those challenges, we still feel very strongly that it forms a very good basis and a jumping off point for the future work, as opposed to going back to the drawing board about the message formats.
So in conclusion, thank you very much for the ability to present. This was very much nuts and bolts, but as we found and we have gone through over the past year a transformation of the financial transaction set by introducing standards, and the pain that has been inflicted, because there are still some holes in those financial standards, we hope that we could put a lot of emphasis on the basics, get the infrastructure in place. Let's make sure we can move messages well back and forth, concentrate on codes that will help us build for the future and offer some of the advanced functionality.
Thank you.
DR. COHN: Thank you all, it has been fascinating testimony. Let's take questions from the committee. Jeff, would you like to start out?
MR. BLAIR: The first thing I want to make is a comment, before I have my question. I would like to applaud what all three of these networks, RxHub and SureScripts and ProxyMed, have done to be able to offer additional efficiencies, improve quality of care and reduce medical errors by the work that you have done. And you have done it in an environment before there were national standards that were really set forth or endorsed by the federal government. There is just a tremendous amount of value to what all three of you as networks have set forth.
I think you can wind up seeing that in the MMA, it is wanting to explore whether there is additional functions that can be included in e-prescribing. Among these I think are making sure that the formularies and the health plans are not restrictive, and that the physician is given an opportunity to respond rapidly and quickly to preauthorizations, to give a preauthorization.
So the genesis of my question, and I guess I have two, one is with respect to the preauthorizations, and the ability of each of your systems to feed back that information to the prescriber, so that if a particular drug is not in the formulary, that the prescriber has an opportunity without any impediments or barriers to come up with an alternative.
The second one is -- and I guess I am morphing into three -- the other one is with respect to medication histories, that the prescriber gets access to that information to be able to make a judgment. Then the third is with respect to what is generally referred to as DUR, but many in the health care provider area seem to call clinical decision support for drug to drug interactions, drug to allergy, drug to lab, that that information gets fed back to the prescribers.
Could you please educate us as to the degree to which this information gets fed back to the prescriber, so that the prescriber has the opportunity to make a modification in the initial prescription according to the prescriber's wishes?
MR. HUTCHINSON: The preauthorization checks in today's world is done primarily electronically at the pharmacy level. In today's world, those calls then, if a preauthorization is needed or required, then those pharmacies are calling the physicians' offices or contacting the physicians' offices in some form or fashion and gaining that preauthorization.
SureScripts as a company is not in the business of the financial transactions or in the preauthorizations for eligibility or benefit transactions. They are more focused on the clinical routing and the prescription information between the physician's office and the pharmacy. So the preauthorization question at this time is probably better for Jim.
MR. BRADLEY: Why don't I try to address all three very quickly, and then we won't be passing the mike back and forth?
In terms of preauthorizations or other kinds of messaging to physicians, RxHub developed something we call a point to point encrypted transaction, in which the physician's office and whoever is trying to authorize the use of a service can exchange messages of any format and so forth. So we didn't see a standard for pre-auth. We developed -- you might think of it as an encrypted envelope, where that could be populated with the information around managing a pre-auth process.
There is a bigger issue here, though. I'm going to go back not necessarily to work done at RxHub, but work done at a startup that -- it was aveton.com, for those of you who remember when I was doing that. We determined that you couldn't be very invasive in the physician's work flow before he took his chosen human interface and threw it against the wall, sometimes harming the patient, and we don't want that to happen. So the first patient safety opportunity is to not disrupt the work flow.
The way we did that is, we took the rules -- and let me just as an example a prescription for a drug in which there are a number of drugs available which might include a preferred brand of drug, a generic, and then some that wouldn't be covered at all. We felt that when the physician was presented with a screen in which they were able to order the drug, that status of each of these drugs ought to be readily apparent.
So if there is a generic alternative, for example, or a preferred brand of drug in the therapeutic category, that is one mouse click or one point that gets to that drug, and if there is then a desire to use an alternative drug, as long as the physician and patient are informed of the consequences of that, which may be total lack of coverage for the drug, could be a higher copay at the pharmacy, whatever. The physician is given the opportunity to order any drug they want, and then to enter information informative to the pre-auth process.
Basically, that shouldn't stop the work flow at all, meaning that that should initiate a process with the payor or whoever is going to authorize it is not invasive to the work flow or practice.
What we found is, most of the time that would facilitate step therapy or other approaches, which would start with the lesser cost drug and work up toward it. The primary principle here though is, there has to be a patient choice for any drug available, and the physician has to have the ability to prescribe any drug available. It may take two or three mouse clicks, but they have to have the ability to do that.
I think your second question was about medication history and how that is made available. In the same scenario, the way we approached -- and again, it is going to differ, depending on the depth of the application that is running in the physician's office, but the way we approached it was, that patient history was downloaded at the point in time that either the appointment was made or the patient checked in for the encounter or the office visit.
It requires patient permission. The way we approached that was, on a blanket basis, the patient signed permission for that download. Then each time they came in for a new office visit, there was a work flow process that got the permission again. That is very important. Then at that point, the patient history was downloaded in a transaction format into the system.
So when the physician chose to prescribe a drug, and I am getting over into the DUR process here a bit, the drug-drug, drug-disease and other kinds of contraindications would be screened against the broad patient history, not just the drugs that this physician knew about because they were prescribed in this practice. That was done. Today we have proposed a transaction format for medication history. We also began to offer medication histories in the inpatient environment. The way we have done it in the inpatient environment is to use HL-7, which is the standard that hospitals know about.
We offer two flavors. One is a text report, the other one is a transaction. The text report, for those hospitals who don't use automated CPOE systems, and the transaction report for those that do. Again, same ten second approach. What it required us to do is actually get inside the firewall of the hospital and basically insert a transaction through the ADT system.
I think I covered all your questions, Jeff.
DR. GUINAN: Yes, the critical factor on the preauthorization is to allow the physician truly to select the drug, to provide information to the physician in the form of decision support. So not to actually force the physician to have to jump through hoops to select the drug that he wants, because then you have to duck in the way of the hand-held device, but rather to present the information in a concise way so that the physician can use it to make his decision. But in all cases, it has to be the physician making the decision.
Now, there are two models that have been deployed in the marketplace for this. One has been, there are a number of databases and companies that have published databases with formularies and preauthorization type information, that gets distributed on CDs out to the physician practice. It gets updated about once a quarter. For the most part it stays correct enough; these formularies are not changed all that often.
That is quite a challenge, distributing these CDs. This is something we participated in for quite awhile, still participate in, as a matter of fact. Our other hat is being a software vendor, besides a network.
This is not really a way to go. Sending out CDs every quarter with this amount of information and worrying about version control out of the physician practices, this is not all that good. So adopting an online transaction that allows for a request response for this would be much preferred.
There is not a standard transaction for this today. There are folks that are working in the marketplace such as RxHub. They are putting forth a transaction for this. But our opinion is that this information will be much better deployed on a broad scale if it was done in a request response type of situation. So it is important to get that transaction standardized.
As far as histories, pretty much the same point. It has been proven that physicians find this very valuable. There is not a standard yet. ProxyMed has participated in sending patient prescription histories to physicians in a proprietary format, based on the script standard, so the basic segments that you needed that were in script, then we concocted our own message format.
There has been quite a bit of work done at NCPDP about this transaction type. This has not been formalized, and it is not part of the official standard. So I believe that also, NCPDP could very well be the jumpoff point.
However, when you speak to the EMR vendors, as a rule they are much more used to using HL-7 to represent medication history in their systems. So it is very important -- and this is going on right now between NCP to PN HL-7. We just need to make a decision. Are we going to go with the fact that most of the information in the EMRs is stored using the HL-7 as the input, and then come up with an interface to script, just make it the HL-7 period and that is what we are going to deploy? Or call it script and then have to worry about interfaces in the physician practice. But our experience has been in dealing with the EMR and truly clinical products for these types of histories, they are used to HL-7 interfaces.
There are many, many interfaces built to the inpatient setting, sending this type of information out to the outpatient setting today. So I would call it at the moment a 50-50 proposition. But that is a decision that needs to be made. Either one could be used. Important to get a decision made so that we can move forward, so the software vendors have -- it takes a few years to get software products built, QA'd, actually deployed out through their release cycles into the physician practices.
As far as DUR, pretty much the same comments as the patient history. There are databases that are deployed via the CD out to the physician office today, from folks like MediSpan and First Data Bank. These work fine, but it is a challenge to be constantly distributed and updating this information out across 200 -- so if you wanted to deploy to every physician practice, this is a challenge. An online type request response transaction wouldn't be much of a burden in drug deployment.
MR. BLAIR: Can I echo back to make sure that I understand this correctly? I think I have heard that standards don't exist for communicating on an online, real time basis either medication history or the formulary information to allow for adjustments to -- if there was not preauthorization, to be able to obtain preauthorization, or DUR information, all of these to the prescriber.
Now, there have been a number of things that have been done because these standards have not been in place to try to accomplish this. It is possible that we don't have to have national standards for all of these. I think that is something that we are going to continue to explore. But I think I have heard from you that standards don't exist in these three areas today.
MR. BRADLEY: I think it is perhaps better news than we may have perhaps led you to believe. In the ambulatory side for medication history, basically that medication history transaction is inheriting a lot from script. So I am presuming that in essence, that will successfully evolve out of the process with NCPDP. I don't think that we are going to have much of a conflict about it.
On the hospital side, on the clinical side, I believe we are going to have to learn to speak HL-7, because the magnitude of the problem to get those clinical systems changed to match our ambulatory view is not going to happen. But again, I think that we are nearly there in terms of being able to pretty clearly define for you what those transactions look like. So I don't think we are in trouble on that one at all.
MR. BECKLEY: When you are talking about, as the prescriber is in the act of prescribing or pre-scribing, as Jim said, you are correct, there is not messaging going on right now. After the fact, the script standard does support the pharmacy communicating back to the prescriber, I need a prior authorization. We have found a DUR. They are not doing it right now. The potential is there for medication history.
But two of the three you mention, again, it is a little bit after the fact, but the pharmacy can't communicate those back to the prescriber. The question is making that proactive.
MR. HUTCHINSON: Two comments. On the prescriber for medication history -- and Jim and I have had lots of discussions about this, we firmly believe there needs to be a level of partnership in the industry around providing medication history. There are medication history databases that exist in the pharmacies, in the payors, in the PBMs as well, that need to be combined. You have cash transactions, you have uninsured.
Of course, you are here today talking about the Medicare and the drug benefit plan, but we all know this is a standard that will be implemented as he described as a whole. So we need to make sure that when we look at medication history, that we look across a multitude of databases, where medication information exists, to be able to combine that.
The good news is, the pharmacies are coming together to say what do we need to be doing. They have started that planning process now of how to provide level of medication history to the physicians upon request.
The final comment, without repeating anything that has already been said around the DUR process is, we firmly believe that the DUR process is not a single stage process. It is a multi-stage process, in the sense that there are -- a DUR is performed at the physician level using technology in these drug databases for drug interaction checking. But you have another stage which occurs at the pharmacy that occurs today at the pharmacy before anything is dispensed against their own drug databases within that pharmacy store, and you have a third stage where that DUR is performed against the payor and PBM database as a part of the claims process. So you really have a multi-stage process with the DUR, so that we don't point to one direction and say, that is where it should be.
MR. REYNOLDS: First, I'd like to say, excellent testimony, and you really summarized it for me. Mr. Bradley, you mentioned the PBM indicator in your patient identifier. Is that just the name of the PBM or is that some kind of an identifying number plus the name of the PBM?
MR. BRADLEY: Let me just give you a little more context on the challenge. The challenge as we started working through this problem at RxHub is, our owners were three major payors or agents of payors that competed with each other. So what we had to do is, we had to come up with a combined structure that didn't challenge the competitive issues that they had amongst each other. So we couldn't store too much. We couldn't become the next evil PBM in the ether or anything like that. We had to really work through a lot of it.
So we wanted to come up with a minimum data set that would allow us to perform the patient identification function. So the five demographic elements identifies the human being reliably. Then we had to store what was a minimum amount of information around, how do we direct to then where the drug coverage is, without replicating that information.
So what we did is, we identified the actual organization, so it would be CareMark, and then the unique identifier which they use to track that individual. That is all we have. It does imply -- and this is the same for medication history and everything else -- is that people who participate in a network like this have to provide real time response to a transaction.
So what happens is, the five demographic items identify Harry Reynolds, your drug coverage is Blue Cross of wherever. We then point at Blue Cross, we have the unique identifier, and we hand off the transaction to them that requests a med history on you. Then they have to have a real time response to us -- we like it in less than three seconds -- where that medication history comes back and we deliver it.
We don't open the envelope. We don't store medication history. We deliver it to your clinician after you have given permission.
MR. REYNOLDS: Mr. Guinan, your discussion about the facility versus the prescriber, do you feel that by staying at the facility level but only including the prescriber in the actual record, will help with secure IDs and the other things that we are all going to have to deal with as we move into the next level, which is, you actually submitted the prescription?
DR. GUINAN: That is correct. To us, they are two very different things. The routing element of how we know where to send the refill or renewal authorization request from the pharmacy is to the facility. Attached inside that message is a piece of information, no different than the drug is, who is the authorizing prescriber that started it.
Likewise for a new prescription when it goes from the physician office through the pharmacy, it is very important to record the facility it came from. Included in that message happens to be, just like the drug or the SIG, the prescriber that authorized this. But it is not the prescriber that dictates the EDI routing of the message.
MR. REYNOLDS: You would say that a facility -- as we move forward, a facility might have one ID with many prescribers underneath it?
DR. GUINAN: That is correct, very similar to the financial transactions in health care for claims, where we have a unique submitter ID that is associated with every service location. In fact, when we submit claims, it is that submitter ID, not the rendering, billing, referring physician IDs. Those are just information in the claim. It is the ID of the location that is all important when it comes to the routing.
MR. REYNOLDS: One last question. The last one is, as you look at the different settings that we have been dealing with, obviously you are dealing -- in the physician's office, we are dealing with somebody that is an inpatient. But a lot of doctors actually visit people in the hospital as they are leaving, write scripts and leave. How do you see in this e-prescribing that -- that is a significant situation that exists. How do you see that occurring in this e-prescribing world?
DR. GUINAN: Well, in terms of just the inpatient to the ambulatory, one of the number-one drivers, we think -- we are at the beginnings of this, but we think one of the number-one drivers of patient medication errors is the absolute lack of continuity between ambulatory care being provided and then what happens in the hospital and then back out again. So that is problem number one, so that has to be addressed.
The thing that is very problematic today is that the hospital tends to choose a formulary based on some form of volume purchasing arrangement. They use those drugs as part of an inpatient episode for which they are at risk, on a per case or per diem or whatever basis. When they go to discharge, most of the time the discharge script is written on the ambulatory drug benefit. So a different formulary applies.
The error potential happens because we believe that patients, before they even see their attending physician before they get home, go back to the drugs that are in their medicine cabinet, not understanding that there is a potential for therapeutic duplication and so forth.
So we believe a couple of things. One is, if there is an ambulatory prescribing process as part of discharge, it needs to be equally aware of the patient's history, benefit rules and so forth, when the discharge scripts are written, because if not, we have the same potential for error. We also have a cost issue. For a Medicare patient, that discharge script would be covered perhaps by a different benefit than the benefit that covers the hospital stay. It is going to be required that Medicare communicate the appropriate formulary and other information at that point in time.
MR. HUTCHINSON: From my perspective, there are a couple of things. I would respectfully disagree with my colleague from ProxyMed on the office based ID versus provider ID. We in fact do implement the provider ID down to that level for the very specific reason that you mention, to allow mobile access for physicians in multiple locations to be able to prescribe medications.
There are many different types of systems out there, and we had to design a capability to accommodate those different types of systems, whether those be mobile devices or whether they be client server applications or web-based applications, or where they may be. It is important to know where those prescriptions are coming from. It also matches up with how the pharmacies look at the information on their side with respect to the provider information within the pharmacy databases.
The challenges on the inpatient on the ambulatory side, however, is the fact that the NCPDP script standard as an example, as mentioned earlier, is not necessarily prevalent as the standard of choice. HL-7 is the standard of choice. Now, with the prescribing vendors, absolutely, the NCPDP script is prevalent within the ambulatory environment. Within the traditional and more established electronic medical record companies that are out there, the HL-7 is the traditional standard that they look at. In many instances, they are not ready to accept the NCPDP script standard.
We are looking right now at a partnership between HL-7 and NCPDP to help us do some level of translation between the HL-7 standards and the NCPDP script standard. There are many, many of the inpatient system vendors that exist out there today that want to send their prescriptions electronically from the hospital setting to the patient's pharmacy of choice, as well as from the emergency rooms to the patient's pharmacy of choice. They are in the process of connecting into the network today.
DR. STEINDEL: First a comment and then a question. I'd like to thank ProxyMed for introducing the routing issue as an issue that needs to be considered and looked at. In light of SureScripts' comments just now, we can see it is a very complex issue. CDC has been trying to raise this issue for a long time I'm glad that it came out in this session. Thank you.
Now to my question. One sense that I got from both your presentations and from reading the written testimony of various people today is questions concerning medication coding. One thing that we are supposed to be doing is identifying standards to use. As many people in this room are aware, last year this subcommittee spent a lot of time looking at standards for medication coding, both through the CHI initiative and our NCVHS PMRI terminology initiative, and recommended a series of standards.
What I am hearing both from your testimony and what I am reading in written testimony is that there appears to be some tension as to exactly what is the most appropriate set of codes to use for ordering drugs. In some cases, we talked about NDCs, because that is actually what is being dispensed and ordered. In some cases we talk about the clinical drug as expressed by Rx Norm. I think my basic question that I have for the group, and then there is a little secondary question concerning the SIG itself, is, at the time these were presented to NCVHS, they were being presented as in development. Do the people in the panel feel that once the system is fully developed, will that meet the needs of the medication coding, or should we be looking at something else?
By fully developed, I mean in essence, if Rx Norm is fully populated within the UMLS and tied into the NDC either as it exists today or as it is being evolved into, or we are hearing that it is being evolved into, will that meet the needs of medication coding, will the introduction of NDFRT physiological access help in the ordering of drugs, and will the structured product labelling as proposed or envisioned by the FDA, which is as I understand coming into effect in probably two phases, first just from a header point of view with text afterwards, but eventually will a fully computable label, would that type of medication system, once it is put into place, will it meet the needs of the e-prescribing as described in the MMA?
Then as a followup to that, it is somewhat corollary to this, because it has to do with coding. It is a very quick one. It is the question of the SIG. Now that SNOMED is relatively available, has any of the groups looked at SNOMED as a source for that information? Because it appears to me that the codes are there, and they could be embedded in any of these terminologies that we are talking about.
So that is somewhat of a followup, I would like you as Jeff pointed out to address the first question as your main point or comment.
MR. HUTCHINSON: I'm going to let my colleague, Ken Whittemore take that, my pharmacist friend.
DR. WHITTEMORE: I'll start in reverse order. I'd say as an organization, we have not spent a lot of time looking at SNOMED, and that is something that we will have to take a look at. So that addresses the SIG issue.
In terms of Rx Norm, we recently became aware of that, and have started to analyze it. I would say -- and this is very tentative in terms of giving a response -- that if there was a connection made through that particular database to the NDC code, then I think probably as an organization we could make that work. But again, it is a little premature for us to -- I don't think we know enough about Rx Norm yet to be able to give a definitive answer.
MR. BRADLEY: Let me make an attempt at this with the caveat that I wasn't smart enough to understand your question fully.
Everything I know is from this group of clinicians and pharmacists and others that worked with me at Abeton and still are alive and well at MacKeston. So I know what they are working on. What I would answer you is, yes, there is a need for a standardization of the ordering process. The challenge is that in the work flow for the physician and in their minds, they don't want to deal with the complexities of NDC coding down to the component level. That is more of an inventory control problem at the pharmacy.
So they think, if I picked up this right, that Rx Norm is the answer to that, that there will be a simplified ordering process. There will still be an NDC coding process when the drug is dispensed at the pharmacy, and the two will be tied. But the physician can prescribe a generic without having to get into who the manufacturer was, can actually delegate a piece of that, with the idea that the closed loop is when the claim comes in to the payor with a very specific NDC code, as to what was dispensed to the patient.
The answer is, they do believe that SNOMED is the answer for the coding of the SIG, period, and are broadly adopting SNOMED within their systems, or at least that is what I hear the clinicians saying. So if the rest of the vendors go that way, game over, that answers the question there.
I hope that was helpful. I can introduce you to the people working on this that know something about it. I know nothing about it.
DR. GUINAN: I'll just add, I am certainly not an expert on this. I would say that the stated goals of Rx Norm and what they are hoping to accomplish is what is needed. The requirement though is an absolute interface to NDC, because there is no way that a pharmacy can operate its business without -- if you want an online order on the front end from a physician, without tying it to their inventory. So the inventory ordering is going to continue to be done through NDC, their claims adjudicated by HIPAA using the NDC code. So if something like Rx Norm is going to be adopted, it has to be interfaced directly to and be able to uniquely identify which NDC is going to be dispensed.
MR. BRADLEY: Just one more comment through Jeff. Dr. Bob Elson, who was with me at Abeton, then MacKesson and now RxHub, and Dr. Kim McCowski, are our experts on this. Jeff, I know you are in dialogue with Bob all the time, so I would pass that on to him and ask him to weigh in on the subject.
MR. BLAIR: Not all the time, but I am very pleased to have him educate us. Anything that I learn from Bob, I pass on to the subcommittee.
MR. BRADLEY: He would be our expert on the subject.
DR. WHITTEMORE: I think the challenge from my EMR days is, yes, a lot of the EMR companies several years ago started implementing the capability for SNOMED, and actually giving provides the option of which coding mechanism they wanted to use. However, the challenge has been physicians' willingness to accept that standard as the standard to be used within those systems.
So I think even if we have it as a standard, from a technical standpoint we still have a hurdle to get over to get physician acceptance.
DR. STEINDEL: This is a bit of a followup. All of you addressed the question of the NDC to the Rx Norm transitions, and I think you did a very nice -- I thank you very much for those comments.
Do you have any comments on the proposed structured labeling and how it would fit into this process, especially when it becomes fully comparable?
PARTICIPANT: Just in general, not for the SIG, the FDA label contains a tremendous amount of information on the drug. Once it becomes understandable by a computer system, I think it has a lot of impact on what we are talking about today.
DR. GUINAN: ProxyMed does not participate in that specific aspect of the process, so we would have no --
DR. STEINDEL: Thank you.
MS. TRUDEL: I am hoping you can address the distinction that I thought that I heard Jim Bradley make between eligibility information and formulary information. It seems to me that MMA considers them to be part of the same concept. I thought I heard you say that the 270-271 appeared to be your choice for eligibility. But then I heard someone else say that there was no standard for formulary. So I guess I am wondering, should they be part and parcel of the same transaction, and how do you feel that they link into the existing script standard?
MR. BRADLEY: I'll start this. The eligibility piece of it, what we are referring to more along the lines of an insurance eligibility, says there is a drug benefit that applies, with some indication of what the drug benefit is, meaning some pointer to a patient specific formulary or preferred drug list or whatever, so the physician system could point at that.
When you get into formulary, the formulary transaction is a bit different. I'm trying to figure out the best way to contrast this. There are two ways that formulary information is made available by most of the technology vendors. One is in real time, in which a transaction -- here is the patient, here is the drug I am choosing to prescribe, is this on the formulary. The problem with that transaction, while real time, is that it forces the physician to shop for a drug. It is very invasive to the work flow. It really slows the work flow process.
So most of the applications that better suit the work flow of the clinic actually download batch formulary loads from whoever the payor is with an indexing structure that allows a pointer. So when you get the patient, you point to that formulary database.
So what we had to develop were two standard transactions, one that had to do with the downloading and batch of these formulary records, the second a real-time transaction that said either this drug is on formulary, not on formulary and so forth. Those are two different transaction standards to support a process, which I agree with you should be combined. This is the drug benefit, here is the formulary pointer that applies at a patient specific level, and then hopefully to aid the work flow in the clinic, we are not shopping, is this drug on formulary, is that drug on formulary, is that drug on formulary. That we are pretty sure doesn't work very well.
Did I answer your question?
MR. BECKLEY: Karen, on your last piece you said what about the script standard. It has the capabilities. It has insurance information in it. There can be a message to check eligibility and bring back formulary at the same time.
Again, you run across issues of speed, how long does a physician want to wait for all that information and come back. But it could be sent in advance. It could be one message. The capabilities are there. It just needs to be implemented and people need to agree to use it. The main points, I'm sure, need some fine tuning, but more than a foundation is there. Quite a bit of the framework is there to do it also.
DR. GUINAN: I will add that ProxyMed is one of the primary providers of physicians getting medical eligibility information back. It is a common practice for offices, hospitals, it doesn't matter what kind of setting to, when they are getting ready to see patients the next day or whatever their process is, pulling charts and so forth, to shoot off a large batch of eligibility requests. That then comes back and they match it up with the schedule, here is all the information we know about the patients.
Well, when we get those back today from the payors, there is everything on there you ever wanted to know except for medication prescription coverage. It seems natural to us, and part of the same process, as they are preparing for patients the next day, it should all be part of this one request. When they do an ANSI 270-271 as mandated by HIPAA, that we should figure out a way to include the prescription coverage eligibility, not formulary.
So the second part of the question, this does not include which drugs are eligible for how the physicians should make that choice, but rather, what sort of coverage do they have, and what kind of rules are going to apply. This is going to get much more complicated, depending on how all the benefits are implemented under the act. If patients can move between various entities that can provide coverage to them, it is very important to know which one they belong to.
So eligibility and the formulary, two different things, and we view it that it should just be part of the 270-271 as part of the normal practice they do today for pulling patient eligibility.
MS. TRUDEL: Thank you.
MR. BRADLEY: One quick followup. Part of the focus more on the insurance industry's traditional standards, we have a coordination of benefits opportunity for the first time. When you enroll this population, there are going to be a number of things we can do to coordinate coverage, with pretty significant implications. But the adjudication of COB requires a certain amount of information that isn't necessarily available in some of the standards that are being tossed around. The process is very complex, and with the high volume of drug transactions, you can't afford human intervention in the COB process.
So my thought is that the 270-271 gets some of that information better and is better able to allow us to adjudicate a broader COB approach that includes drugs. That is part of the reason we went there.
DR. COHN: Could one of you explain to me a little bit more about in Medicare Part D, how it is going to work with COB? I hadn't really -- my understanding of the Part D benefit is that COB will probably be a minor if any piece of this. Can you clarify?
MR. BRADLEY: Yes. I worry about the commercial world. Whatever you adopt will close over.
The other thing is, let me just recount the first time I met Mark McClellan. I was starting to explain to him about the payors' issues with electronic prescribing. He respectfully stopped me and said, look, I understand all that. I want to talk about the direct restructure. We think that over time, coordination of benefits is going to be an important part of a public-private partnership with how Medicare drugs are administered.
So to the extent that private risk parties and other insurance vehicles interoperate to deliver a drug coverage to a senior, there may well be a COB opportunity with Part D. So again, I am thinking about the future and what it brings.
DR. COHN: Thank you.
DR. HUFF: Since everybody has been talking about HL-y, I have to disclose that I have a bias for HL-7. I am the former chair of HL-7, currently a vocabulary co-chair in HL-7. So put that in context.
My question relates to the kind of clinical information that -- we have talked variously about medication history and quote, medical history, that would be transmitted. It is not 100 percent clear to me what the scope of that should be and when it is flowing, whether it is flowing from the physician to pharmacies or physicians to benefit managers. Could you clarify what your vision is of what -- is it purely just medication history, or do you want lab data and diagnostic information, that kind of information? When doe it flow and who does it flow to?
MR. HUTCHINSON: The intent is what is available today and what is going on down the road or in the future, as Jim represents, that we need to be focused on. Today, obviously HL-7 is very critical for lab result delivery into EMAR systems as well as into electronic prescribing applications. That is the default standard used in the ambulatory care environment for the exchange of clinical information even between systems, whether between a hospital system and an ambulatory setting system or a lab system in the ambulatory setting system. So I would agree with you. In that sense HL-7 is the default standard in the ambulatory care environment.
Around medication history, in today's world of what is being implemented, it is the payor to PBM delivery of medication history information to the physician. In a broad scale manner today, in the DUR process that occurs in the pharmacy, it is not using an HL-7 standard. The medication information is delivered in the form of an alert during the process of a DUR alert. However, the history information itself is not delivered directly to the pharmacies, which we would like to see that history information delivered to both care providers, the physician and the pharmacy. I think that has to be an industry partnership between the pharmacies, the payors and the PBMs to be able to pull together a complete medication history for patients.
MR. BRADLEY: The context of the medication history that we are talking about, again this dates back, was first implemented in Cocomo, Indiana in 1999 with Medco. There, the encounter in the physician's office was enhanced by a downloading of the medication history from the payor database. So this downloading occurred between the payor and the physician's office and was stored as part of a record there.
The breadth of that was again with the patient's permission, all physicians prescribing, all pharmacies dispensing. It is still today. Each time a physician is introduced to it, they feel that it is data that they have not had before.
Now, in terms of the broadening of that, we have been approached by one national lab company to say, now that you can identify the patient in a quarter of a second, we believe that we could participate in such an information routing as well. So the idea that lab values could flow through the same pipe.
The problem on laboratory though is, if you take the large national labs, for example, and put their combined business together, I think it is 30 percent, maybe 35 percent, of the world. An awful lot of things still happen in the hospital, happen in the physician's office, or happen in regional lab companies.
The thing that has made medication history effective for us thus far is the significant consolidation of the PBM industry. Three pipes get you to 75, 80 percent of the commercially insured population. So therefore, physicians are willing to try the transaction in an attempt to get the medication history, because most of the time we can deliver it. We aren't there with some of these other forms of data yet.
DR. GUINAN: I'll add that the HL-7 is the primary transport format of lab data amongst physician practices, hospitals and so forth. Today, the script is the primary format of the messages for e-prescribing. It doesn't seem that to have a physician's office receive both lab results information as well as medical records, the two have to be. The EMR companies and practice management systems have some well-established HL-7 interfaces. The script standard is very easy to implement.
It has not been a challenge to get folks that already have HL-7 interfaces to implement script for this data stream coming in. The data is not stored in HL-7 or script once it gets there; it is stored in whatever internal representation in their database.
So my point being, it doesn't seem that there has to be the same transport protocol for lab and prescription. It would be very nice if through one request, the physician could go out to something in the sky and say, give me all the information about this one patient, but that might be asking too much.
So in the meantime, if payors are able to distribute medication histories via script and lab results come from the labs via HL-7, it just does not seem to be an impediment for it to all end up in the same EMR in the physician's office. So my point being, I don't think we need to move in the direction to try to have to marry the two together. They both can stand on their own.
DR. WHITTEMORE: If I could just add one professional issue, as we mentioned earlier, SureScripts was created by a Community Pharmacy. Community Pharmacy for quite awhile has felt that their ability to counsel patients properly as they are required to do so by law and board regulations, that would be enhanced if they could get laboratory information and diagnosis information. So as the committee contemplates this kind of messaging back and forth, I would certainly hope that that would be part of the process that it will consider.
DR. HUFF: In that regard, I guess the thing that seems hard is, the messaging is actually pretty well understood. We know how to send messages for lab and/or diagnostic information. It is that -- to me, it is how many copies do we need. We have the information, the diagnostic and lab information eventually ends up in the EMR in the physician's office, and then it gets copied to the pharmacist and then to the payors, and how many copies of this data do we need, and how do I recognize duplicates of the data if it is requested again.
It now looks like the pharmacies are maintaining an EMR that is essentially a duplicate of what is in the -- is there a boundary on that? Is there some way to manage and regulate, or do we just do it?
DR. WHITTEMORE: I think clearly there would have to be some kind of a boundary. How far would you go down the road in terms of an EMR, in terms of what information a pharmacy would need in order to do what they do. I would think that would be part of a dialogue that would take place in the standards setting organization.
MR. HUTCHINSON: But I would add to that, that unlike my examples previously on the banking industry, where we may use one bank, in the health care world, as you know, patients are all around in multiple places in health care. I think you are going to have multiple copies of information in different databases, because a patient is in multiple health care organizations, whether that be family physician or a cardiologist or a hospital or a pharmacy or other areas. It is a matter of what level of information is needed at which points of care to improve the safety and efficiency of the process.
The inefficiencies that occur around this today in the pharmacies is having to call the physician's office to make sure a lab test has been performed, to find out what the results are, and to handle that with the payor or PBM to be able to dispense the medication, and have to clear up that prior to. If it is delivered electronically, it reduces another call, and that is obviously a much safer transaction as well.
DR. STEINDEL: Over the last couple of years, this subcommittee has heard numerous testimony from deliverers of alternative medicine. I am recalling Jim Bradley's distribution slide on assembling a medical history from various components. In no place in that did I see a direct link to alternative medicine providers, if they weren't in a pharmacy or something like that. Does this pane l have any comments on how this subcommittee should consider that?
MR. BRADLEY: I'm not sure there is an infrastructure to arrive at that information, other than one thing that we haven't talked about this morning very much, the patient and their role in fleshing out what this medication history looks like.
So one thought being that the patient needs to also contribute. In the absence of any other sort of information, it would be nice to know about aspirin therapy, it would be nice to know about other alternative medications, vitamins, that the patient may be taking.
So I think there is an opportunity. I get asked this about once a week. There are a number of organizations that are advocating now for a patient record, patient controlled, that can be contributed as part of the information base here. We are seeing a movement within the health insurance industry towards more consumer directed care, to make them more accountable for how they use the health care system. So I think these things work hand in hand. We need to figure that out.
The issue with seniors, I'll just give you an example. My mother last week said, I can't understand these discount cards, can you help me. I said, the reason that they exist is to make you more sensitive to the cost of providing care to you and to provide you with some mechanism for relief. She goes, but I don't understand. Then I realized I couldn't help her.
So anyway, my point being that seniors are sometimes confused. We need to help them help us. The woman I am talking about here prevented my father from a very, very serious medication error recently, because of lack of communication between the physician and the pharmacist, that didn't occur, and almost caused a double dosing of a very significant medication.
DR. GUINAN: I'll just add that I think it was actually an omission that we did not talk at all about patients when we were speaking before. In fact, when you talk about providing medication histories to physicians, the patient will want to know what is that physician seeing.
This has been one of the challenges in getting that service deployed, has been getting the authorization from the patients, because before you can get all pharmacies, all payors, all medications to the physician, the patient has to say, this is okay for this doctor to see these other medications. Although physicians of course want to see this, they said it is very valuable, the patients are going, wait a second, I'm not so sure I want this physician to see what is going somewhere else, as well as coming from a family that is quite a big user of alternative medicines, I don't think my mother wants her physicians to know all of the alternative medicines, because her primary physician doesn't like this and yells at her about these things.
So patients absolutely are going to want to see what is in the content of those medication histories that are being passed around. That just leads further to more of a patient centric model, and the patients controlling what is in the content of the medication histories, which is a whole other can of worms that get opened up, but might be unavoidable.
DR. SCANLON: It seems like all three of the networks that the panel described this morning are regional, interstate and even national operations. Are there any state legal or regulatory barriers that remain in terms of wider scale in prescribing, or have you more or less dealt with those, and there are no further ones to bring to the panel's attention?
DR. WHITTEMORE: I will take that one, because I am spending and have spent probably the last half a year working on this.
The practice of prescribing and dispensing medications is regulated primarily at the state level, with the exception obviously of controlled substances, which are governed by the
Drug Enforcement Administration. Early on in the life of our organization, we took a cursory look at the regulations at the state level, and our conclusion was that it appeared at that time that electronic prescribing was allowed in the various states by the regulations.
We got to the implementation stage, and actually got to the point where we were rolling that out, and we found that we probably didn't ask the right question at the beginning. What you really have to take a look at when you look at these regulations, is electronic prescribing permitted, the way that you do it.
That is what has taken our organization close to a year to work through with the various states. You run into situations where the regulations just don't recognize electronic prescribing as a legitimate way of transmitting a prescription. Some of the regulations have been in existence, as Jack said, this is going back six or seven or eight years. The boards of pharmacy, they had regulations back then regulating something that they didn't really understand yet, and so many times the regulations don't fit the technology.
There is an issue that we run into with some regularity with respect to electronic versus digital signatures that has to be cleared up, depending on what the regulations say in a state. So there are a number of remaining issues. In specific, there is language that we find quite a lot around the states. I'll paraphrase it, but it usually runs something to the effect that an electronic prescribing must be transmitted to the pharmacy of the patient's choice, with no intervening third party entity having access to the prescription, and then they all embellish on exactly how they say that. Depending on who is doing the interpretation and looking at that dictates whether you can operate in that state or not.
So we have spent a lot of time. As an organization we basically have cleared our process in 31 states now. We have got about another ten in process. But it has taken a long time. It is again very organization specific. So I think there is an impediment that the committee needs to look at here.
The good part of the story is that the National Association of Boards of Pharmacy, who I believe you are probably going to hear from at your next meeting, have changed their thinking a little bit in terms of the role electronic prescribing can play in terms of patient safety. They now as an organization are advocates of eliminating handwritten prescriptions. I believe you will be hearing from them and their efforts to change the regulations. So basically what would happen is, they would come up with changes to their Model Pharmacy Practice Act, which would disseminate out among the states.
DR. COHN: As the Chair, I am going to take the prerogative to wrap up at this point on the session. I want to thank our presenters. It has been a fascinating and hopeful session for all of the members of the subcommittee and staff.
I am reminded that this is an area that is in the crosshairs or nexus between a lot of the different standards activities and a lot of other emerging pieces. We obviously as a subcommittee need to do what we can to create simplicity and clarity in this area. Many of you have been moving forward and making decisions based on what you had to deal with at the time, but hopefully we can provide some guidance for you, we hope, in the near future.
So anyway, thank you very much. We will take a break for 15 minutes and come back at 11:30.
(Brief recess.)
DR. COHN: We are pleased to welcome Doug Bell. He will be speaking about a study that you have done and some of the recommendations from it. So thank you for joining us.
DR. BELL: Thank you. I would like to thank the subcommittee for inviting me here to summarize work we have done at Rand to develop recommendations for electronic prescribing systems. The main result of this work is the actual recommendation that resulted from an expert panel process we conducted. But I am also going to show you some preliminary results from a field study that we have done to look at what is going on among commercially available e-prescribing systems in comparison with the recommendations.
Just to tell you a little bit more about me before I start, I am a practicing internist and faculty member at UCLA in general medicine, in addition to being at Rand. I have a Ph.D in health services research and I also have training in medical informatics.
Why did we undertake this project to create e-prescribing recommendations? We conceived this about three years ago when we saw the explosion of new products coming into the marketplace, e-prescribing products. That was creating a situation that we still have, where there were a lot of different options available for electronic prescribing.
The systems fulfill a variety of laudable aims, like improving work flow in the office and increasing formulary adherence. But our goal was to make sure the patients' interests were represented in the e-prescribing systems. In general terms, we saw patients' interests of course being patient safety, meaning reducing medication errors, and in a broader sense, health outcomes which would mean getting the right medications when a patient needs particular medications, then also helping patients manage their costs. We were especially sensitive to that, because a few studies at Rand which have just come out in the last few months have shown that patients are especially sensitive to cost changes. Often they don't have a chance to negotiate decisions that they are making based on costs to their providers. So as the prior speakers have said, patients get to the pharmacy, they are surprised by the costs and often fail to take medications that they really need.
So with these goals, we set out to create recommendations for e-prescribing. We were fortunate to have funding from Pfizer to do this study.
Again, the specific objectives of the study were to create recommendations for electronic prescribing, to promote patients' best interests, which I just summarized, including safety, health and cost management, and also that don't hinder e-prescribing adoption or violate patient privacy, and that are supported by a rigorous objective multidisciplinary process. So that was the first main objective. Also, we wanted to assess how often commercial electronic prescribing systems are already implementing the recommendations that the panel was going to come up with.
The methods were, first we did a literature review, and that produced a summary of evidence. We fed that into a delphi expert panel process. That produced the set of recommendations, which were published today actually as a web exclusive in Health Affairs. I believe everybody has that document. Then we conducted a field study which has produced an assessment of current systems in comparison with the recommendations.
Here are the results of the literature review. This slide summarizes our results. First, notice that we didn't treat electronic prescribing as a black box, which a lot of evaluation studies might. We wanted to look at the specific features of electronic prescribing and look at what effects might be anticipated from those specific features.
We published the results of the literature review in a paper in Journal of the American Medical Informatics Association this year. I wanted that to go out to the panel, but I'm not sure it actually has. But we will try to make sure that that gets to everybody.
I don't want to spend time going through each individual finding that we had here, but suffice it to say that the effects we found varied quite a bit from quite small effects to some quite large effects. But overall, the literature was weak. There were lots of potentially important features for which there was no evidence, and even the evidence that does exist has uncertain generalizability at best. Most of these studies are conducted in very unique environments, and there is often only one or two studies. So overall, the literature is quite limited on this.
Also, the focus in the literature has been primarily on medication errors, and not very much if at all on overuse or under use of medications. In some other recent studies that we have done at Rand, we are finding that under use of medications is potentially a bigger problem than medication errors, in terms of the lost potential for improving patients' health.
There is a study that was just published in Annals two or three weeks ago from Rand's ACOV study, assessing the care of vulnerable elders, where they found that only three percent of vulnerable elderly patients were getting potentially harmful medications, whereas 50 percent of the time when a patient had a specific need for a highly effective medication, and they were actually not getting that medication. So there is some evidence emerging that under use of highly effective medications is as big a problem as medication errors, even in the elderly.
We have found that even when there is a lot of evidence in the field at Rand, that to develop specific recommendations, we often still need expert opinion to be able to interpret the evidence that exists. In this field, we really needed to go beyond what the evidence was, and to synthesize expert opinion to create recommendations.
So we convened an expert panel. We started by recruiting Don Berwick, who is head of the Institute for Health Care Improvement, as our chair. He helped us recruit a group of other recognized experts across these different domains. I won't go through each of these people. They are listed in the Health Affairs paper in the acknowledgements at the end.
Now, because e-prescribing was such a new field, we asked the panel to participate in drafting the initial recommendations. We held an initial panel meeting, where we presented a literature review, and then conducted a nominal group process in which we went around the room and asked each panelist to name their top ideas for patient centered recommendations.
The project staff then organized the dozens of suggestions that came out into categories, and they also reworded them to make them explicitly describe one desirable e-prescribing capability in a consistent way, so we have consistently worded recommendations.
Then we took these recommendations through three rounds of revisions with the panel. We collected the panel's feedback using some preliminary ratings in each round, teleconferences, and we collected written comments. We extensively revised the wording. We really focused on the meaning of individual words in these recommendations. We also drafted new recommendations, added recommendations to the set as we went.
So the final set we ended up with was 60 recommendations. We took that into the delphi expert panel process with more formal ratings. We asked the panel to rate each recommendation's effect on four dimensions: patient safety and health outcomes, helping patients manage costs, maintaining patient privacy, and promoting clinician acceptance. Of course, our main interest was in health outcomes and also in costs, and we were just hoping to not damage privacy or clinician acceptance.
The ratings scale that they used ranged from minus seven to plus seven, where zero represented no effect, essentially the panel saying this recommendation would have no effect on this particular scale. A rating in the lower third of the scale indicated a clearly negative effect, and to use a rating in the upper third of the scale from plus three or above to indicate a clearly positive effect.
One other thing we asked them to do is to rate how soon each recommendation would be achievable in the average clinician's office. We actually provided them with preliminary results of our field study. It was still in data collection at the time, and we had a rolling process of looking at what was going on in the field. However, they were asked to consider not just what was technically achievable, but what is achievable in the average clinician's office.
The complete results of our study are in the online data supplement to the Health Affairs paper. This is what it looks like, just to give you a sense of that. I think to really get a gut sense of what we did, I do think it is important to take a look at this. The full wording of each recommendation is here. I'm sorry you guys don't have this, but we will try to get it for you.
There is also a rationale statement for each recommendation. That was present as the panel did these ratings. These are the median ratings on each dimension for each of the recommendations. We just drilled this down to whether it was achievable within three years or not. So this is what that looks like, and I just wanted to show that to you so you can look for it later.
This slide shows how the 60 recommendations were organized into ten categories. The first seven categories shown in white here correspond to the steps a prescriber would follow in caring for an individual patient. The latter three categories are broader recommendations that apply across caring for individual patients. The category that had the most was medication selection, but you can take a look at these again later as you go through individual ones.
I wanted to walk through just two sample recommendations in detail, to give you a sense of what we did. This is recommendation number seven, in the category of access to patient historical data. It says, prescribers with care responsibility for the patient should be able to review the patient's complete current medication list, based on open prescriptions from all other clinicians.
This was carefully worded. The wording was honed. In fact, we added in the bit about providers with care responsibility for the patient. This was something that emerged during the panel discussions. That might help to limit the privacy implications of having a complete medication list. But overall, the median rating was at the very highest end of the scale for safety and health for this recommendation. It also was just into the positive range for patients' costs. It did have slightly negative privacy implications, but it didn't meet the threshold for clearly negative effect on privacy. Then to our surprise to some degree, it was rated as being strongly positive for clinician acceptance. Also, the median panel result for achievability was that this would be achievable within three years.
In the medication selection category, recommendation 21 says, the system should enable providers to determine the accurate formulary status, and the actual cost to the patient for each medication option, based on the patient's prescription insurance coverage. This was our main formulary recommendation.
This didn't quite make it into the clearly positive range for patient safety and health, although we did extensively discuss the potential health implications of knowing the cost at the time of prescribing. The panel didn't quite rate it high enough.
It had the highest possible rating for helping patients manage their costs, no effect on patient privacy, and a weakly positive effect on clinician acceptance, according to the median panel ratings. They did not think that this was achievable within three years, because we were asking not only for the accurate formulary status, but also for medications that are not in the formulary, we thought that there should be some way of estimating the patient's actual costs. They thought that integrating that information together might not be achievable.
Just stepping back, this histogram shows how median ratings came out on the patient safety and health dimension for all 60 of the recommendations. Fifty-two of the recommendations ended up in the really positive range, with plus three or above, and eight did not make it into that range, but all of them were at least in the positive direction. None of them were in the negative direction.
On helping patients manage their costs, 18 of the recommendations were in the clearly positive range, with the rest mostly being more neutral. Note of course that there is quite a bit of overlap between these 18 recommendations. I don't have a specific slide on that, but obviously quite a few recommendations were both clearly positive for safety and health outcomes, and also for costs. So each of these recommendations can have multiple purposes.
On the clinician acceptance dimension, 55 were rated in the clearly positive range, with only a couple being on the verge of being significantly negative.
Finally, for maintaining patient privacy, almost all of them were rated as having no impact, with a few that were specifically aimed at privacy having fairly high ratings on that dimension.
Now, note then that none of the recommendations were rated as clearly negative on any dimension. So overall, we had no feedback from this group of experts that these recommendations would cause physicians to revolt against electronic prescribing, or that they would significantly invade patient privacy.
Finally, it is not shown on a graphic, but 43 of the recommendations were rated as achievable as of three years, and the other 17 were rated as not achievable within three years.
So that concludes my overview of the results, and we can discuss it more, of course. But I also wanted to give you preliminary results of our field study. We wanted to assess the extent to which these recommendations were being implemented, and were really for the range of what is going on in the marketplace. These results are still preliminary because they are under review at Health Affairs actually right now. So as the review process goes, there could be changes in emphases. We can't publicly release the full results yet, but I do have permission to talk about these results right now from Health Affairs.
We began by trying to identify all of the companies that might produce e-prescribing systems from a variety of sources. We identified 129 companies that we screened to see whether they actually produced an e-prescribing product. We were able to identify 58 different e-prescribing products from 51 companies. Then we had inclusion criteria that were -- we wanted to focus on outpatient systems and systems that had significant market adoption, which we defined as having at least a thousand users or being installed at 50 sites or more. We just went on the vendor claims at that point, of whether they had that much adoption or not.
MR. BLAIR: Just for clarification, when you say companies, are you excluding at this particular time the networks, like from our previous panel? Are you just talking about the software developers that produce the e-prescribing systems for the use of prescribers?
DR. BELL: That's right. We were focused on what are the end products at a physician's office or a prescriber. We tried to avoid saying physician, because there are lots of prescribers who are not physicians; what a prescriber would actually use, what is the end product, essentially. Some of our recommendations do focus on lower level underpinnings of e-prescribing systems, but we wanted to focus on what is observable in the clinical environment. So we didn't look at those specifically, although we did try to make sure we included companies that worked with those intermediaries, and we did actually -- ProxyMed was actually one of the companies that we did select for site visits. We promised these companies that we would not identify -- you will see the results; we are not going to tell you -- break the code and tell you who is who, but they did all agree to be identified as a group, and ProxyMed did happen to be one of the ten we selected.
So we selected ten companies to do site visits, to go out, look at how the systems were being used, and assess in person, is each of the recommendations being met. Again, I said representative on the slide, but we are not trying to do something that would be statistically representative of what is going on in the marketplace, because we realize that that is impossible with the rapidity with which the marketplace is evolving. We were interested in what is the range of what is going on.
We used a purposive sampling method to try to at least represent some products from different types of systems, and here is what we ended up with. This shows you the types of systems. We wanted to include both electronic medical record systems and those that are completely independent, stand-alone e-prescribing systems. So we had five electronic medical record products and five that were not, that were stand-alone e-prescribing.
We also wanted to include hand-helds, so five of the systems were available in a hand-held platform, and the other five were not available on a hand-held platform; they had to use a desktop. Then we wanted to include the newer web-based application service providers. We ended up including four of those. The other six were locally installed client server systems, rather than web-based application service providers. As I mentioned, we did want to make sure we included some that worked with the intermediary companies, and we did.
Here is the main result. On average, the systems that we site visited had implemented 50 percent of the standards. It is actually 49.7 percent. This graphic shows the implementation level for each of the individual products that we looked at. Each product is lettered from A to J, and the height of the bar shows the proportion of the recommendations that were either fully implemented or partially implemented. Some of the recommendations had multiple parts, and it was possible to partially implement some of the recommendations.
These first five are the EMR systems, and the last five are the ones that were non-EMR based systems. So system A had fully implemented 63 percent of the recommendations and partially implemented another 14 percent, whereas system J had implemented 26 percent of the recommendations fully, and then partially implemented another 16 percent of the recommendations.
The main thing to note here is the degree of variance among the systems. These all again were systems that had achieved a reasonable level of adoption within the marketplace. One interpretation of this variance is that the marketplace isn't creating a convergence, at least yet, on patient centered features that we have in our recommendations.
DR. COHN: Doug, before you move from this slide, I am presuming that each of the features were equally rated?
DR. BELL: That's right. We didn't try to do a weighting based on the ratings, which would be another thing to potentially do. But this is just the raw proportion implemented.
I just wanted to briefly focus on implementation of a subset of 15 out of the 60 that we thought had something to do with addressing under use. As I mentioned before, there was very little literature on features addressing under use as opposed to medication errors, so we thought that this is an area that deserves more emphasis.
This is eight out of 15, just to give you a flavor. These are all paraphrased, actually. The exact wording is more detailed than this. This is the completely current medication list recommendation that I showed you, and five out of the ten systems fully implemented that feature at the time we site visited. By the way, the site visits were conducted in 2002 and 2003. They were over by August 2003, so things I'm sure have evolved some since then. But five systems at that time were doing complete medication lists.
Another recommendation was to provide diagnosis based medication menus. That means you could enter a diagnosis and then get up a menu of medications that might be indicated for that. Five out of the ten systems were doing that. We have a recommendation that says prescribers should be notified when prescriptions are not filled within some reasonable time interval that the provider could specify. None of the systems were doing that, giving alerts for non-adherence.
We have a recommendation, and this is really an interoperability recommendation also. It says systems should extract data for decision support from external sources, including pharmacy, hospital, laboratory and EMR systems. None of the systems were fully implementing that, meaning that they did all four of those -- they had all four of those external interfaces, although some systems were doing a lot of individual interfaces with different types of outside systems.
We have a recommendation that says there should be reminders for putting a lid on recommendations, so this is very much an under use recommendation. Only one system was really trying to do that.
Rather than going through all the rest of these, I'll just say that overall, for these 15 recommendations, 37 percent were implemented among the group of systems. So a somewhat lower level of implementation of the under use recommendations compared with the set of recommendations as a whole.
Just to conclude, here are the main take-home messages I want to make. First our expert panel produced 60 recommendations that could be used for creating federal e-prescribing standards. The dimensions we used to rate these mapped to the goals that are laid out in the Medicare Modernization Act.
Those four goals, I'm sure you are familiar with, are to improve safety, quality, efficiency and interoperability of prescribing systems. Our patient safety and health outcome scale essentially combined these two together, because we see the medication errors and the safety as one part of improving patients' health in the long run.
Our patients' cost dimension is a patient centered aspect of efficiency. We didn't have an interoperability scale per se, but a lot of our recommendations were focused on interoperability and on enhancing interoperability. We saw it more as a means to an end than an end in itself. So these ratings could help you choose among the recommendations as you create standards.
We also found in our field study that commercially available systems don't consistently implement important features. So standards are needed to insure that at least minimum features are implemented in the marketplace. We are demonstrating how important the work of this subcommittee is in establishing the bounds in which e-prescribing systems will compete.
Finally, our results showed that more study is needed in areas where evidence is limited. This is relevant to the subcommittee, because the Medicare Modernization Act calls for an evaluation of federal standards in 2006, and this could be structured to address some of the unanswered questions in the e-prescribing area.
Just a few to emphasize that emerged from our discussions, one of course is how e-prescribing could be used to reduce under use. I guess I have already talked that through enough, but we need more evidence in that area.
Another issue that came up was whether transparency is enough to deter biases that could theoretically be introduced by third party sponsors. We pretty much took the position that transparency in disclosure would be sufficient. We didn't find any evidence that it wouldn't be, but some of the reviewers of our paper felt that especially any third party sponsorship of e-prescribing should be banned.
Obviously, that is not consistent with what is in the Medicare Modernization Act, and we didn't think that there was evidence for it, but as we move forward in this field, it might be worthwhile to have some kind of audit mechanism built in to e-prescribing so that people could look for potential biases more easily.
Another question is how much room systems should have for human judgment. This is an area of disagreement among the panelists, with some panelists believing that systems are generally more fallible than humans, and allowing things like alert overrides could actually be a bigger loophole for providers to make mistakes and to do the wrong thing than the converse, which would be that you have to trust human judgment in the end. So we think that more research into how overrides get used could be something that is important to answer that question.
Finally, e-prescribing will be a major socio technical change. Any socio technical change can have unforeseen consequences, and we think it is going to be important to monitor this area for unintended consequences as we move forward.
Finally, this is an area where there is going to be a lot of change in the next ten years. It would probably be useful to set up a way for standards to be regularly updated.
So that is the conclusion of my prepared remarks.
DR. COHN: Thank you very much, interesting presentation. DR. STEINDEL: Yes, thank you, it was a very interesting presentation. I have a question concerning the use of the word standard as you used it and the way we have been interpreting it from the Medicare Modernization Act.
The way I interpreted the way you used standard is, you came up with these 60 or so recommendations, which I would look at as attributes of e-prescribing systems. Then you referred to them as standards that should be implemented in e-prescribing systems. We have been looking at standards as the nuts and bolts of doing this step or that step, the implementation of these attributes and standards in that way.
Now, in the HL-7 process, they are in the process of producing a draft standard for trial use, which will probably be published in July for the electronic health record. Many of the attributes that you have put out just on my quick reading are attributes that may appear in that EHR standard. Have you done any reconciliation between what your recommendations are and what is being recommended with the electronic health record system, in part so that we can make sure that both systems, when they go into place, move forward with the same set of attributes?
DR. BELL: Yes. Thanks, that is a great question. First of all, in terms of standards, these were intended to be non-technical standards. So they are somewhat different than part of your charge, which is clearly to do the nuts and bolts of how these systems hook up. But we felt that at a higher level, some non-technical standards might also be useful to establish bounds for what -- actually, not only bounds for competition among e-prescribing systems, but also aspirational capabilities that these companies can reach for. We really were hoping for both.
I occasionally use the word standards to describe what we did, because when we originally did the study, we were calling them standards or potential standards. But as we went through the review process, clearly they should be recommendations, not standards. We don't have the evidence base or the authority to really call these standards at all. So they really are recommendations, what we are issuing, and not standards.
The other parts of your question were -- I have not done a careful comparison of these with what has emerged from the electronic health --
DR. STEINDEL: HL-7 EHR.
DR. BELL: The HL-7 EHR. There was also a set of standards recently issued.
DR. COHN: A set of recommendations.
DR. BELL: From the EHI, right, and those are also probably at a more detailed level. I think what we have here is also probably at a less detailed level and a more non-technical level than what HL-7 is doing as well. How these might play out into recommendations is something that we should still work on, I think.
Then your last question was about electronic health records, EMRs, essentially, and what the interface is. Clearly there is a lot of overlap. We saw e-prescribing as a subset of EMR functionality that some companies are breaking off into a separate product, and how well that really flies remains to be seen, I think.
MR. BLAIR: Before I ask my question, let me just elaborate just slightly on a couple of those references, because I think there are a lot of folks in the room that are trying to follow this, and may not be familiar with some of the references there.
When the reference there was to HL-7, the HL-7 EHR functionality model work that has just been done included functions that included e-prescribing and a number of other related functions that would deal with the prescribing functions, whether they were decisions support functions or eligibility functions or whatever. So that was the reference on HL-7.
On the EHI, the E-Health Initiative has just come out with a very significant report on e-prescribing, and one of its chapters includes recommendations on standards, health care informatics standards that would facilitate e-prescribing. So those of you that are checking it out, that is what that was.
My question. I think that the report and the study that you did, I would hope, might be able to help us in trying to address one of the following problems that we have to deal with.
Let me describe the problem, and then maybe you could tell us whether there are recommendations coming out of here that could help us with this.
Electronic prescribing initiatives, including the networks that we just heard from, have gone forward in a pragmatic fashion to try to solve problems, and they have done an outstanding job with what is available now. They have made investments, and they are getting a lot of acceptance.
The law winds up asking this committee to look at e-prescribing in a very broad and deep manner, where we are considering a lot of standards that would enable functions that would improve patient safety and quality of care and costs that go well beyond what is being used in the industry today.
One of the things that I am wondering from your study, from the way that you described it, is, if it could help us with trying to understand where the points of diminishing return would be as we begin to look at what additional future standards may be accepted by the market in a particular time frame.
I'll just give you one example, and then maybe that is enough for you to wind up telling me if your study sheds light on these types of things, okay?
DR. BELL: Okay.
MR. BLAIR: The law winds up asking us for a lot of drug to allergy, drug to lab, drug to even patients' weight and all, and related to safety. It also asks for patient history, which you would think would be in the EHR.
Now, a lot of those standards are not in place, but they are being done in other ways, either in proprietary manners, which may be fine. Did your study reflect on where there might be pushback by physicians in terms of the time that it would take -- this is the essence of the question. It may take a prescriber more time to do all of the checking, to make sure that things are really safe, high quality.
Is there some indication of where the threshold is for how much we could ask a prescriber to do in checking out all of these things, even though they all contribute to safety, but at what point do we get diminishing returns?
DR. BELL: Sure. You are asking about the tradeoff between adoption and the potential safety benefits. Each of these features may require some alteration of what doctors do on an everyday basis, and if there is too much, there will be a revolt. It will hinder adoption.
We do have ratings for both the safety impact and the clinician adoption impact of a variety of recommendations that do relate to some of the things you just suggested. For instance, weight-based dosing. We have a recommendation on that, and I can't tell you off the top of my head how the ratings came out, but we would have to look at somehow combining the ratings on each of those two different dimensions to look at which ones might come out as the highest priority, if you wanted to weight those two dimensions most strongly.
We probably haven't thought of everything that you might be interested in and that the law might have even asked for, but we do have some depth among the recommendations we have. But I guess ultimately, there needs to be an empirical question as well.
DR. COHN: Maybe I should pipe in here a little bit. I want to first of all thank you for the presentation. I think it is fascinating. It is good to get some additional evidence about what is there now versus what is three years, versus what is a gleam in a legislator's eye.
DR. BELL: I wish we could release the results of the field study right now too, but hopefully it will be soon.
DR. COHN: I'm sure it will be more interesting than all of that. I guess no one having seen the people participating in the delphi study, obviously the strengths of these activities are that you get real experts with great insight. But on the other hand, they have real experts in certain areas, and I would suspect just looking at the membership, probably the area where they would be least capable of making an accurate determination might actually be physician acceptance, because many of them are not actively using these things on a day in and day out basis.
DR. BELL: That's right. We did over represent physicians, I think, but you're right. We relied on Clem McDonald a lot, actually. He was our informaticist, and he was extremely --
DR. COHN: We have relied on him a lot in our committee at times.
DR. BELL: In terms of getting views expressed to the other panelists.
DR. COHN: And certainly he continues to practice actively. I was just observing that I think I would be looking at your report to focus around some of the quality views. I thought the three-year view of what is realistic in three years was a very important view. I think the field study when we know more about it will be a little bit sobering, in terms of realizing that our reality is not as far as our vision on this. It may help us a lot.
But certainly, Dr. McDonald ably represents a lot of our views in a lot of places. I don't know if you realize he used to be on this committee, and just recently, maybe six months ago, left this committee. So we miss him.
Anyway, are there any final questions before we break for lunch at this point? Judy, you haven't asked a question before.
DR. WARREN: I'm just wondering, as I have been sitting here listening to the networks presenting and then listening to your study, everyone talks about clinician work flow. I am sitting here worrying about how are they going to accept some of this new technology, and interested in that.
Has anybody taken a look at faculty and how they know about work flow, and how are they going to be teaching our future students about the necessary work flow that is going to be required through using technology?
DR. BELL: I know there is some awareness and desire to teach that, if you just look at the discussion that goes on in the medical education lists and in that literature. But in terms of actual evidence or studies to look at that, I'm not aware of any.
DR. WARREN: And nothing came up in your discussion among the panel members?
DR. BELL: We actually didn't focus on the educational environment. It was really focusing on the community practitioner.
DR. COHN: Harry, we will allow you one question and then we will break for lunch.
MR. REYNOLDS: Thank you, nice job. I really enjoyed it. You mentioned -- and I would like to ask the question only for illustrative purposes, so you won't be indicting anyone. What are the types of influences that you see that concerned you? Where were you guys coming from?
DR. BELL: One of the reviewers of our paper was extremely worried that third party sponsors could bury some kind of bias deep within decision support rules. There certainly were no examples given, and we haven't found any examples, but they were concerned that they could be hidden away in ways that would bias the prescribing process, and that would not be readily observable.
In the absence of specific evidence, I guess our position was -- and we have transparency and accountability recommendations. If you look, there is actually a paragraph toward the end of the paper that specifically talks about the transparency and accountability recommendations. I could tick a few of them off for you if you want, but we felt that on the input side, that the sponsorship needed to be fully disclosed. Also, the decision support rules themselves also needed to be available for inspection essentially, not that the end provider would ever do that, the end user, but that somebody should be able to inspect the decision support rules themselves. That sort of combination of disclosure would go a long way to deterring biases.
One area where we do have a recommendation that there be a prohibition is on specific promotional considerations being used to alter the display of medication options. So if a PBM or a pharma was interested in promoting their brand of a medication and they paid a vendor to show something with a smiley face next to it or in gold print, that was something that we said should not be done. That is recommendation 16, I think. But beyond that, we were aligned mostly on transparency.
MR. REYNOLDS: That is helpful.
DR. COHN: Well, Doug, I want to thank you for your presentation. At this point we will be adjourned for lunch. We will come back at 1:15 and continue the conversation.
(The meeting recessed for lunch at 12:26 p.m., to reconvene at 1:24 p.m.)
A F T E R N O O N S E S S I O N (1:24 p.m.)
DR. COHN: We will get started with this afternoon's session. We are pleased to have David Bates joining us to talk about patient safety in e-prescribing. David, thank you for coming.
DR. BATES: Thank you very much for having me. It is a pleasure to be here today. I think the committee's work is going to be very important.
I am going to start by talking a little bit about the epidemiology of the problem. Though I know the main focus is the outpatient setting, I'll say a few words about medication safety in the inpatient setting, because that is the area that we know the most about. Then I'll talk about the outpatient setting, very briefly about nursing homes. I'll talk about some of the evidence that computerizing prescribing improves safety. I'll spend some minutes exploring issues relating to decision support, which I think will be a nice followup to some of the things that Doug Bell raised, and then I will conclude.
One study that we did that looked at the magnitude of the problem of medication errors and adverse events in the inpatient setting was the adverse drug event prevention study. This was done in two large hospitals in Massachusetts. The key findings were that there were 6.5 adverse drug events for every 100 admissions. About a third of them were preventable. There were three potential adverse drug events or near misses for every preventable adverse drug event, and 62 percent of the errors were at the ordering and transcription stages. Most of the research up to this point had focused on the drug administration and dispensing, and this emphasized the importance of prescribing.
Another study we did evaluated this problem looked at medication errors, and the chance that they would actually result in an injury, in a study that was published in the Journal of General Internal Medicine in 1995. This study was done in three units, two general care units and one intensive care unit. In 10,000 orders roughly, there were 530 medication errors, 35 potential adverse drug events, and five preventable adverse drug events. In this study, about one in 100 medication errors resulted in an injury or an adverse drug event, and seven in 100 had the potential to do so.
More recently in evaluating the issue of medication safety in the outpatient setting -- and there is, as I said before, less evidence about the magnitude of the problem in this setting. In one study, we surveyed over 2,000 outpatients in a variety of clinics in the Boston area. Eighteen percent of patients reported a problem with their medication, 35 percent of the time after they did so, their medication was not changed. Twenty percent of patients had symptoms that lasted longer than three months. Three percent had adverse drug events when we subsequently did chart reviews. Of those patients, five percent required hospitalization, 13 percent of those adverse drug events were preventable.
Clearly in this study, there was a discrepancy between what the patients were reporting, and what we found in the charts, 18 percent versus three percent. We were not able to determine in this study what the true frequency was, although we got some information from the next study that I will tell you about.
Additional findings from the study that I just described were that these problems did have an important impact on both patients and the system. So they were strongly associated with lower patient satisfaction. Forty-eight percent of patients sought medical attention because of these issues, 49 percent reported experiencing worry or discomfort. Correlates of patient reported drug complications were number of medical problems, failure to explain side effects on the part of providers in the primary language of the patient other than English or Spanish.
Another study that we did that was published in the New England Journal of Medicine this last year was a study called the IMP study, the improving medication prescribing study. That study included 629 patients in the Boston area, outpatients. Of these patients, 132 experienced non-preventable adverse drug events for a rate of 20.9 for every 100 patients, 74 had a preventable adverse drug event for a rate of 11.8 for every 100 patients. Among the 206 adverse drug events, 86 percent were significant, which is the lowest level of severity, 14 percent were serious, 36 percent were preventable, and six percent were both serious and preventable.
We also evaluated the frequency of adverse drug events in nursing homes. This work has been led by Jerry Gerwitz, who is at the University of Massachusetts. In one study which included 20 nursing homes in Eastern and Central Massachusetts, there were 546 adverse drug events, or 1.9 for every 100 resident months. There were 276 preventable adverse drug events, for a rate of about one per 100 resident months. Seventy-two percent of the serious events were preventable, 47 percent of the errors were at the ordering stage and 53 percent were at the monitoring stage. That is quite a bit different than what we see at the inpatient setting, in which monitoring is not nearly so important; in nursing homes, monitoring is very important. Psychoactive drugs and anticoagulants were the classes of medications most commonly implicated in the preventable adverse drug events.
Admissions due to adverse drug events are also important. I just wanted to emphasize this, because patients are particularly vulnerable when they transition, either come into the hospital or go home. There are relatively few recent data about the magnitude of this issue. The literature reports a wide range of admissions being related to adverse drug events, between a half a percent and 21 percent of all admissions.
One study that we did recently found that 1.4 percent of admissions to our hospital were due to adverse drug events. All of these originated in the outpatient setting, 78 percent were judged to be severe, and 28 percent were preventable. So this represents an important opportunity for improving safety.
Another study that we did recently looked at the incidence and severity of adverse events after discharge. This study included 400 medical inpatients in the Boston area. In that study we called patients just after they went home, and to our surprise, the adverse event rate was quite high at 19 percent. Six percent of these were preventable, six percent were ameliorable, which was a new term that we coined, which meant that we couldn't have prevented the adverse event, but we felt that we could have made it substantially less severe had we basically been in touch with the patient earlier. Of the adverse events, 66 percent or two thirds were adverse drug events, and 17 percent were procedure related.
This is an example of a preventable adverse event. A patient with congestive heart failure was started on spironolactone. They were already on an ace inhibitor, Lasix and potassium. There is a good randomized control trial showing that in the aggregate, this combination of drugs is beneficial. But this patient, unlike in the trials, didn't have any electrolyte monitoring, and the clinicians in the rom will be able to predict what happened. The patient became extremely weak and anorexic in two weeks, and their potassium was over 7.5. This is clearly something that could have been prevented were different systems in place.
This is an example of handwriting. I think people in the room are familiar with this sort of thing. Any guesses as to what this says? This is a prescription for Abandia, which is an oral hypoglycemic, and not an anticoagulant at all. Handwriting isn't the biggest problem in safety that we have, it is not the only problem we have. We find that typically it only represents about 10 percent of errors, but it is certainly a problem that we should be able to deal with today.
So if you computerize prescribing, it improves the quality of drug ordering in a number of ways. First of all, it streamlines and structures the process, so when people are picking doses from menus, they won't choose a dose that is ten times too high. You can decrease or even eliminate transcription. You can require complete orders. It turns out that many prescription orders when they are handwritten do not include a dose or a route or frequency. With the computer, you can make those required fields and make sure that they all make sense.
You can also give people information at the time that they need it, so you can show relevant laboratory tests, you can display guidelines and put in place guided dose algorithms. Then perhaps most importantly, you can perform a number of checks in the background, for example, look for drug allergies, drug interactions, put in place dose ceilings, do drug-patient characteristic checking and do drug lab checking.
I had thought when I first got involved in this area that we would do the things that are down there at the bottom, and that I would go on to do different things. That was many years ago at this point. We are still learning about how best to provide decision support in each of those areas, and learning about which decision support to provide. I will give examples of that later on.
I think this group actually could contribute a great deal if it could make recommendations about how to represent some of these pieces of information. I'll go through in detail about why I believe that is important.
One study we did, which was a randomized control trial, which used both contemporaneous and time series comparisons, we evaluated the impact of a very basic computerized prescribing application. It required that all orders be completed in minimized transcription, and it included some very early checking, including checking for the most important drug allergies, penicillin allergies and sulpha allergies, but not a comprehensive array of allergy detection at all, and the most important drug-drug interactions, but again, not a comprehensive array of drug-drug interactions, so just about the top ten drug-drug interactions, and no other decision support whatsoever. Despite the limited decision support, we still saw 55 percent decrease in the serious medication error rate. This was in the inpatient setting.
This is another study that we did in the inpatient setting, in which we looked at the effect of computerized prescribing on the medication error rate. This was all medication errors and not the serious medication errors. The serious medication errors are the ones with potential for harm. But the overall medication error rate fell from 140 per thousand patient days, about 68 percent after computerized prescribing in period one. It bumped up a little bit in period two. In period three it had decreased over 80 percent.
Another study that looked at the impact of computerized decision support was a study called the Nephros study, which was published in JAMA in 2001. In this study, we evaluated the effect of real time decision support for patients with renal insufficiency.
The way that this was done is, the computer knows the patient's age, their gender. We asked the clinicians to provide the weight, and then the computer did a calculation and suggested the appropriate dose and frequency of a medication based on those data.
In this study of around 18,000 patients, 42 percent, which is many more than we had expected, had some degree of renal insufficiency. In the control group, patients only received the appropriate dose 54 percent of the time, they only received the appropriate frequency 35 percent of the time. In the intervention group, that went up to 67 percent and 59 percent, which was a substantial improvement, and patients stayed in the hospital a half a day less, which was a statistically significant decrease.
I sometimes get asked why those figures in the intervention group are not much higher, why it isn't 95 percent. I think that is an important question. We do find that in most instances, physicians do elect to ignore even good recommendations a significant proportion of the time. I think we probably need to make stronger suggestions in this sort of situation, although we are not right 100 percent of the time, and I will give some more examples later about how often we are incorrect. I think it is important to let clinicians override decision support.
In the outpatient setting, we have just completed a study focused on improving outpatient prescribing. I don't yet have the results to share with you today, but I will tell you what we did.
In the first year of this study, this study was done jointly with Regenstrief, which is at the University of Indiana, in the first year of the study, we built a real-time adverse drug event monitor, so we could use the computer system basically to determine how often patients were having adverse drug events. In the second year, what we did was to implement basic computerized prescribing. So we had allergy checking, drug-drug interaction checking, but not a comprehensive array of checking. Then in the third year, what we were doing was evaluating the impact of complex decision support, so we have added on top of that renal checking, hepatic dose checking, checks regarding pregnancy and guided dosing algorithms.
This doesn't map directly, I suppose, to the legislation, but in some ways it comes close in many regards. It was actually quite challenging to determine what decision support should be provided down here at the bottom. We actually had to develop a lot of that on our own. Even though there are some suggestions available from the literature, much of that was not readily available.
Findings that emerged from the IMP study were that basically, of the intervention strategies, computerizing prescribing appeared to be the highest yield strategy for improving patient safety in the outpatient setting, and allergy detection appeared to be particularly important. Notification about drug-drug interactions also appeared to be particularly important. Dose suggestions were the third most important category.
It also was clear that better tools for followup are pivotal. Patients had trouble contacting practices. Many patients were experiencing problems and did not let anyone know about the problems they were experiencing.
In nursing homes, we have nearly completed a trial of computerizing prescribing. This has been done in two nursing homes. One is in Connecticut, another one is in Toronto. We picked these nursing homes because they were the only two that we could find in North America that had computerized prescribing. Overall, the level of information technology in nursing homes is quite low, and this is an area that clearly needs attention, although I think there is enormous opportunity for improvement here.
I believe that key areas of decision support with respect to medication safety, and the key areas that we need to work on, are the following. This is based on a series of studies that we have done.
The first of these areas is just requiring complete prescriptions. That is not a challenging thing to do. If you can get someone to prescribe electronically, it is relatively straightforward to require that a prescription be complete.
The next highest yield thing to do is actually to suggest default doses. Again, that is not enormously challenging from the IT perspective, but it turns out to be very valuable, because when people are shown a range of options that are possibly correct, they are much more likely to pick the correct or safe option.
The third highest yield area based on some analyses we have done is probably based on renal dosing. That is not what I would have expected at the beginning, but it appears to be a very important area. Again, it is an area that is not that difficult to approach once people begin prescribing using the computer.
Drug allergy checking is the next highest yield area. I will be saying quite a bit more about that. That actually is a complicated area, and it is one that I believe we need more standards around, and I will describe why.
Drug-drug interaction checking also turns out to be important. The biggest issue there is that we don't have very good knowledge about which interactions to warn providers about. Many of the commercially available products have too low a threshold with respect to warning clinicians, so there is a real risk of driving clinicians crazy by giving them too many warnings. That has sometimes very substantially impeded adoption. So it is important to determine what level of interaction to alert about.
Geriatric dosing is another area which I believe is important. We have done a study looking at the impact of this in the inpatient setting. That is under review now, so I can't tell you what the results are. But it stands to reason that it would be useful to select the right dose of medication, given a patient's age. There are many instances in which older patients should receive lower starting doses of medications than younger patients.
Drug laboratory checking is also an important area. There are many instances in which a particular laboratory level suggests that a medication should not be used, or it should be used at a different dose. That is a very high yield area.
Dose ceilings is another important area. For most medications there is a certain dose that you should not exceed. That is again not terribly challenging from the informatics perspective.
Next I would like to talk some more about allergy checking. We recently did an evaluation of the way that we were doing allergy checking at our own organization. Partners Health Care is a large integrated delivery system that includes two major hospitals and a number of other hospitals. We have both inpatient and outpatient clinical applications. So we just looked to see how we were doing things just within our own organization.
What we found was that drug allergy checking was remarkably inconsistent across the clinical applications. We actually weren't doing things the same way in any two of the individual applications, and the differences had important implications for safety.
We did not have a common allergy database. So if you came and had an allergic reaction in one part of our system, we didn't -- if you had one at the Brigham, we did not move it over to the Mass General, or vice versa. That is just one obvious example. All the applications handled allergies differently, and I'll go through some of the ways in which that was the case.
We also found that most drug allergy warnings were being overridden. About 80 percent were being overridden, and that proportion had climbed substantially over the last several years. An additional issue was that when patients had new allergic reactions, those often did not get entered. Obviously the computer can't warn the clinician if the information doesn't get entered. Only 16 percent of new allergic reactions were entered in one study.
Some of the key features of our former drug allergy checking application were entry of the allergy and the reaction. All the applications included that. The allergen was coded in most instances, but the reaction might have been or might not have been. Actually, in many instances it was not coded. That had important implications for decision support, because you wanted to deliver different decision support if the patient had an anaphylactic reaction, as opposed to nausea and vomiting.
Drug allergy interaction checking was handled in different ways. Sometimes, some of our applications included cross-sensitivity checking, and some did not. There were issues with groups. So for example, ampicillin and penicillin are obviously in the same family. If the patient is allergic to ampicillin and they get an order for penicillin, there should be a warning, and vice versa.
Reverse allergy checking was only variably present. In this instance, the issue is checking pre-existing medications when a new allergy is entered. Then overriding a drug allergy warning, there was variability. An override reason was sometimes required and sometimes not required.
Between 1996 and 2000, we saw a 2.5 fold increase in the number of our drug allergy alerts, but at the same time the acceptance of the alerts fell from 52 percent to 17 percent. We found when we further examined this that 65 percent of the drug allergy alerts were given for the same drug to the same patient. In other words, the clinician had already seen that alert, and had elected to override it, and we still continued to display it.
It also became clear that some cross-sensitivity alerts were probably excessive. in particular, for sulfa drugs and loop diuretics, there is at least a theoretical risk of cross-sensitivity. But when we displayed those warnings, they were virtually always overridden.
The other issue was with narcotics. Often someone will have a problem with one narcotic, and if you give them another narcotic, that is virtually always overridden.
We also felt that some allergy group alerts were probably excessive. It was also clear that some allergies are probably actually intolerances. So if someone gets a stomach upset, that is not a true allergy. Overall, we had the strong impression that the total number of drug allergy alerts that we were delivering was approaching the bothersome level.
Here is an actual example of the way we do this. Here the allergen is erythromycin. We put in a place for a coded allergy here. The reaction is hives. Here is another example of a warning. Here, the drug is trimethafrome sulfamethoxisol. The reaction is irritability. The patient has a definite sensitivity to TMP sulfa.
We are experimenting now with displaying warnings in different ways. It is clear from other industries that if you for example display a warning like this and provide some visual clues, that is probably helpful, in terms of getting clinicians to pay attention to the truly important warnings.
Here is an example of a really important warning. This patient has been ordered for nafcillin intravenously, and they have a prior allergy to penicillins, which was anaphylaxis. We found in the drug-drug interactions that clinicians were just as likely to override the potentially lethal interactions as the ones that were quite minor. So that does suggest that we need to do a better job in terms of cuing providers about the importance of individuals in individual situations.
Future directions in this area. We believe we need to determine better when to alert, but perhaps more importantly, how to alert, so that providers recognize the truly important warnings. There are cognitive issues to deal with and work flow issues.
It is also clearly important to sort out how to get people to document better. We believe overall that it is very important to make it easy to do the right thing, and that we need standards for representation of reactions. That proves to be very important in delivering subsequent decision support, and that is one of the things for which I don't think we have good standards.
I think that we need to also determine what kind of information will prove most useful to clinicians, so we need better information about usability. We need to be doing processing in the background, and understand what the context is that the clinician is in when they are doing a particular activity.
Down the road, we need to work on how to deliver much more complex decision support than this. After all, an allergic situation is relatively simple compared to a lot of the other things that one could imagine doing.
This is a prioritized summary of our specific recommendations after doing this analysis with respect to allergies. It is clear to us that we need to have consistent decision support across applications. It would obviously be nice if there were some sort of consistency across applications in general with respect to the way allergies are represented.
It was clear to us that we needed to reduce unnecessary alerting, and we are in the process of removing certain low yield cross-sensitivities and removing certain group related alerting. It will be nice if every health care organization in the country didn't need to learn the same painful lessons over and over again.
It was clear that we needed to adjust notification based on severity. We need to improve our allergy documentation and also encourage incorrect allergies to be deleted. When we set things up initially, we made it impossible to delete any allergy, and that seemed like a good idea at the time, but it proved to not be very useful. It was also clear to us that we needed to include drug food allergies.
Since we have done this, we have actually implemented a common allergy checking across Partners. Now if you do have an allergy any one place in Partners, that information does move with you. We are doing our allergy checking in the same way. But that proved to be quite difficult, even in one relatively little integrated delivery system.
Another issue that I wanted to mention relates to the inpatient setting. We have been doing some work looking at intravenous medications and how they are delivered, and have identified some interesting and I think important variability which has major safety implications.
It is clear that most computer order entry applications don't use dose limits. That has a problem in and of itself. We have been studying smart pumps, which are pumps that know what medication is being administered, and if the nurse administers too high a level, we can warn the nurse. But to put in those warnings, we needed to have agreement about what the maximal doses should be for certain drugs.
What we found when we convened a multidisciplinary group was that there was very wide variation among institutions in dose limits. Most of them had never compared notes about this particular topic. There was little or no rationale for very major differences in maximal dosages, and there were lots of opportunities for error as a result.
In general, there was tremendous variation for infusion therapy, in terms of drug names -- I will give some examples of that -- concentrations per drug, and dosing units.
Here are a couple of examples. For calcium gluconate, in different institutions, calcium gluconate was represented as grams per hour, milliequivalents per hour, milliequivalents per minute, milligrams per hour, kilogram per hour and milligrams per minute.
Now, medically there is not a lot of rationale for doing it all those different ways. When a nurse goes from one place to another, they may or may not know what the standard is for that unit. Six of one of these things might be fine, but six in another place might be a lethal dose. There is not really any clinical rationale for this type of variability. There is even more variation for magnesium sulfate. We found this type of variability across the board. Even within single institutions, there was often very substantial dosing unit variability, drug name variability and so on.
I just want to mention a couple of things which are related. I believe that beginning to use electronic health records or prescribing might be the biggest single step toward improving outpatient safety, based on many other things that have been said today.
The first thing is pretty clearly computerization of prescribing, but as has also been mentioned, exchange of drug information will be very important for improving safety. It is critical that people be on the same page regarding what medications are actually being delivered.
I wanted to mention that I am chairing a group called National Alliance for Primary Care Informatics, which is a coalition of many of the major primary care societies in this country. That group has come together to try and advocate for things that will increase the use of electronic health records in the outpatient setting, and of course, e-prescribing is a very key part of that.
There are also a number of very interesting things going on at the state level in a number of states, and I am sure you will hear about a number of those. In Massachusetts we have been working with a multi-stakeholder group to try and build a coalition to incent providers to begin using electronic health records, with the end goal of improving safety and medication safety in particular.
So to wrap up, I believe that computerization of prescribing can substantially decrease serious medication error rates. It can probably decrease adverse drug event rates, although the data on that are actually quite limited. The studies that have been done have not been adequately powered to detect a difference in adverse drug event rates, and so that is why we don't have that information. It would cost millions of dollars to do studies like that. Although that is something the country might consider investing in, it is not something that we have done so far.
We do still have much more information for the inpatient settings than for other settings, although increasingly there are more and more data about medication safety outside the hospital. Most of the benefit from computerizing prescribing relates to the associated decision support. What I believe this group can do is to help with standards for representation of things like alerts, reactions and other areas related to medications. I recognize that much of that may not happen in the next year or two, but I think if we are really going to achieve the benefit that we would all like to see with computerization of prescribing, we will need to have standards for some of these areas, or we won't actually get to where we would like to go. Overall, I think that greater standardization for medication related domains has the potential to dramatically improve medication safety because the benefit does relate to the associated decision support.
I'll stop there and take questions.
DR. COHN: Thank you very much. Questions from the subcommittee? Maybe I'll start with one. I was struck, listening to your discussion of this issue of alerts, and I have myself experienced at times, when you get so many alerts you want to do damage to your information system infrastructure. I have also heard at least anecdotal reports that there are pharmacists and others who become overwhelmed by the nature of the pharmacy alerts in some of these systems.
While it is clearly not a standards issue, or maybe it is, I am struggling with how one solves this particular problem. It is clearly right up there high in terms of patient safety, and something we need to somehow make work better. But what are your thoughts on this one?
DR. BATES: I have a few thoughts. One, it is very important which alerts you actually deliver. There is not a lot of reason from my perspective to deliver different drug-drug interaction alerts in every health care system in this country. That is something that, if we could agree about in some way, we wouldn't necessarily have to have a set of standards for which drug-drug interactions to warn on. But if we had an agreed-upon set that we decided as a country were important, and that all vendors could access, that is the kind of thing that would be a resource that would be widely used, and it would help clinicians a lot.
I find that the way things go today, I prescribe something, there is an interaction, I pay attention to what my computer system has suggested. Then the patient gets to the pharmacy and they call me up immediately and say, there is this drug-drug interaction, what do you want to do now. It is redundant.
Often there is checking going on at several levels. If we had better interoperability clearly, some of that could go away. It also seems like there would be a lot of advantages to agreeing about which things to warn about. Doing that, it seems to me, will be hard to do if we have not agreed about how to represent things like rules, and if we are not representing rules in a standard way. I see drug-drug interactions as an example of a rule.
DR. WARREN: When you say a standard representation of a rule, what kind of standard would that look like? Just help me understand more what you are asking.
DR. BATES: There are a variety of standards already for rules, but it would include things like which elements the information system should be looking for, like for example two different drugs, what level of notification is involved, so is this something for which the provider should be interrupted immediately, what type of acknowledgement is required. So there are a number of dimensions that are associated with many rules.
DR. WARREN: You have helped me understand my confusion, at least. When we talk about a set of standards for rules, I think of things like Arden syntax. But when you tell me drugs to look at and stuff, now we are talking more about a knowledge base.
DR. BATES: The two interact, but I am thinking more about things like Arden syntax. I also think that as a country, we need to come to some agreement about what knowledge to deliver, but that is really a separate question.
DR. COHN: I do have one other comment. The other note that I made -- my handwriting looks like your drug prescription example -- but you talked about standard representation of reactions. Have you in your environment come up with what you think is a good standard for that? It seems like maybe SNOMED is the right thing to be using for that, but I don't know if that quite solves the problem.
DR. BATES: I actually don't know. We developed something that we developed ourselves, as we so often do in this sort of situation. I don't know whether SNOMED for example has taken that on, and someone else in the room might know. But clearly this is an area that to deliver good clinical care, we do need a standard for. There are many others as you go through things with medications that are like this.
DR. COHN: Other questions from the subcommittee? Any questions from the audience? Well, David, thank you very much, we appreciate it. I think we are due now for a 15-minute break, so why don't we do that, and we will get started at 1:50.
(Brief recess.)
DR. COHN: Will everyone please be seated? We are going to get started here momentarily. Despite our relatively long break, the good news is that we are still on time, so we are doing well today.
Our next session is hearing advice and guidance from the implementers. I want to thank you all for coming and joining us. We will start with Robin Thomashauer. She and I were talking; I guess I should disclose to everyone that she used to work for Kaiser Permanente many years ago. She is now at the Council for Portable Quality Health Care. We are happy to have you here, and it is good to see you again.
MR. BLAIR: Maybe everyone could introduce themselves first, so that everybody could know what the panel is.
DR. COHN: Sure. Robin, and go across, and introduce yourselves.
MS. THOMASHAUER: I am Robin Thomashauer. I am Executive Director of the Council for Portable Quality Health Care.
DR. MEDVEDEFF: I'm David Medvedeff. I am Vice President, Government Business for Gold Standard Multimedia. We are a company that is implementing the Florida Medicaid Project.
DR. MANDEL: I am Robert Mandel. I am Vice President for E-Health at Blue Cross Blue Shield of Massachusetts.
MS. ADAMS: I am Laurie Adams. I am the President and CEO of the Rhode Island Quality Institute.
DR. COHN: Thank you all for joining us. Robin, would you like to start out?
MS. THOMASHAUER: Thank you for the opportunity to speak to you this afternoon. CAQH is a not-for-profit alliance of leading health plans and networks working together in two primary areas. One is to standardize non-proprietary business processes across health plans to make it easier for providers to work with the plans, and also we have initiatives in the quality arena to support physicians in the delivery of quality care.
You can see from this slide on member organizations who we are working with. We have a cross section of national and regional players, which gives us a pretty broad perspective on the plan's operating systems.
MR. BLAIR: Robin, just for my benefit, maybe you could just mention a few of them.
MS. THOMASHAUER: Aetna, Wellpoint, Anthem, Blue Cross Blue Shield North Carolina, Empire Health in New York, Horizon Blue Cross Blue Shield, Oxford Health Care, Healthnet. Do you get the picture?
MR. BLAIR: I get the idea. Great list.
MS. THOMASHAUER: We have some PPOs at First Health and Multi-Plan, so we do have a broad range of products represented.
There is a broad range of institutions they are working on, but one of the current initiatives under way is an e-prescribing initiative that we began several years ago, as we looked at what opportunities there were for us to look across the plans.
In 2002, we designed a database to integrate new formularies from all of our member plans in standard format in order to provide a single resource for technology vendors and other parties to access formulary information. At that time, there were several vendors who had this information, but it wasn't being sourced by the plans, it wasn't as accurate as it needed to be, and it certainly wasn't as timely. They would get a variety of different mechanisms and the updates were not often included.
So the plans felt, what better source of this information than the plans themselves. Based on the experience we have had in this area, we think we have a unique perspective to share with you today on our experiences involved in the e-prescribing world. The focus of my remarks is going to be on the challenges that we face and the solutions that we have developed to address some of the concerns and issues that would have created barriers to the adoption of e-prescribing.
As we looked in administrative simplification, it became clear that the processes that were being used to write and fill prescriptions were resulting in a great deal of wasted time by the providers, by the patients, and by the pharmacies. In addition, the consumers were not realizing the full benefits of the coverage that was available to them. A good part of this was due to the fact that the information was not available at the point of care, at the point of prescribing.
Part of the problem was the way the information was available. It was available in paper and books that come out from the plans to the physicians. It was also coming out from organizations that were trying to integrate, and who were creating booklets which had several plans available, but again, it was outdated and often inaccurate by the time it got to the physician.
The health plans also have websites available, but most providers in most markets cannot just access that information from one website. Unless the plan has dominant share in the market, they had to toggle back and forth between different plans. In addition to the multiple websites, there were multiple formats and multiple classification systems being used. So the information is very difficult for the physician to access in a way that was meaningful in their work flow processes.
Finally, there is a huge increase in the complexity of the pharmacy benefit, and that information available at the point of care and point of prescribing is increasingly difficult to make available. So those are the kinds of things that we found to be the biggest challenges in moving forward in e-prescribing.
The health plans were pretty clear about the benefits of e-prescribing. The value, the quality of care, patient safety, all of these things are things we know about and I am not going to go into them, but the issue for the plans was, how can we impact that, how can we help in those areas. I think it was clear that writing the formulary information at that point of writing the prescription was going to be the most valuable way that the plans could provide assistance.
In order to get that done, we faced quite a few barriers. The evolving e-prescribing market -- I'm sure you know a lot more about this than I do, but in the last several years we have certainly seen that technology come a long way. The functionality and the applications have really developed exponentially in the last three years, but certainly some of those vendors who are around today were not around then. So as we tried to address that market, the players kept changing and the game kept changing, which made it more and more difficult for us to figure out how we could interact with them in a meaningful way.
In addition to its lack of interoperability which is certainly a barrier for us, the different components of the process -- we start with the physician and enter the pharmacy, and the pieces throughout that process are not talking to each other very effectively, if at all. So that would be the other big barrier that we face.
The early e-prescribing efforts that we will probably be hearing about in the next three days, inconsistency in data format, which was a big problem that we thought we might be able to help, inaccurate and limited data which we discussed already, lack of communication among the stakeholders, and also and most importantly really, was the perception by the vendors that the physicians were not interested in having that formulary information available at the point of care. The physicians did not see that as their role, and as a result, the vendors were reticent to talk to us for quite some time, because they didn't want to develop functionality that would not be of interest to the physician, and thereby the physician wouldn't be interested in the product.
So as we created our initiatives, our two objectives were to simplify the formulary process for physicians and patients, and to help them understand how to manage the health plan formulary process.
The formulary data source was developed to resolve inconsistencies, insure the data was accurate and timely, and assist the providers in making the best treatment decisions. The database was created for the vendors, in addition to which, we realized that another way to get the information out there was on the web. We were not under the illusion that any physician would take the time in their practice to walk over and log on and look at what was out there, but it might be a help to office staff. Also, in those markets where there were dominant vendors, if they were integrated into our database, it might be an opportunity for them to check the formulary quickly, because we were going to standardize the format.
So we did create a free searchable website that was available to anybody interested, and was used by members, patients and providers. Integrating this information required that we standardize data elements and formats where they were displayed. So to do this, we worked with a variety of stakeholders, we worked with plans, we also worked with vendors, and we worked with medical specialty organizations to get their input on what would make the most sense.
We created the standard for the plans to apply to the formats and the data fields in which they submitted the data. We established a quality assessment process for the plans to check that data before it was submitted, and therefore we provided the vendors with accurate formularies that were easy to access and to use.
I would like to acknowledge the collaboration that we accomplished with all of our member plans and medical specialty organizations and others. As I am sure you know, working across these organizations to get standardization is one of the great opportunities that we have in life, so i really wanted to acknowledge them, because they really committed themselves to making this happen. Particularly our formulary work group, which are the pharmacy directors and the PBM directors and the information system experts in our member plans, who are a wonderful resource and could certainly be a help to this subcommittee as you move forward.
We standardized in four particular areas the process. We created an Internet based process whereby the health plans could submit to us. They would collect and verify their information, and then submit if they chose; they could also do it through other means. But they could do this through the Internet, submit their information to us, and quality check those formulary modes. We also developed standard processes for the vendors to receive the formularies from us.
The data content. The data fields that were provided by health plans and requested by vendors varied. So to integrate the data, we developed a series of standardized elements that health plans must use when submitting the data. You can see up there, the drug name at the NDC 11 level, whether or not it is on or off formulary, brand or generic, and modifiers, quantity and age as two examples. Vendor feedback was really essential to us in understanding what needs they had and what they were going to be using in their products, so that this database would be useful to them.
The final format. Each health plan compiled and displayed the formulary in a different format -- Word, Access, Excel and a host of other vendor based solutions, and we standardized which ones we could accept the information in. We also standardized the formats by which we could send the data to the vendors.
The data would be submitted to us. We would standardize the data across all the plans and integrate them, and then provide it to the vendors in the format that they preferred. The standardization clearly allowed an effective way for plans to communicate with the vendors, and also enable the vendors to access to a far greater number of plans' formularies at one time. Before that, the vendors were working one on one with each of the health plans and working with their various formats and fields.
MR. BLAIR: Robin, just a clarification. When you use the word vendors, what vendors in what settings are you sending this information?
MS. THOMASHAUER: Our target vendor audience were the e-prescribing vendors. So whether they be the hand=held vendors or whether they be practice management vendors, they have an e-prescribing component to their functionality, or any software vendor who is creating an e-prescribing component that could potentially create functionality for a formulary bounce. So it was a pretty wide range of vendors that we were talking to.
Lastly, the therapeutic classification drug. Different plans used different classification systems, primarily First Data Bank and MediSpan. In order to build this database, we consolidated or integrated the two into a third classification system that built on those two. We built a crosswalk so we could put either one of those classification systems or formularies that were in these classification systems into this third classification system that we could then use as our standard.
More industry wide discussion certainly needs to occur in this area. The standards for therapeutic classification can help improve the flow of classification by eliminating the need for translation between all the parties, which as you might imagine took us quite some time to accomplish.
There are significant issues in doing this. There are huge resources that have already been invested in creating the e-prescribing tools using these current standards, which means extensive efforts are required to reprogram. I think that certainly the clinical experts who develop these need to be engaged, but in our experience, we really feel strongly that a standard therapeutic classification system needs to be developed to take out a lot of the waste and the time that is required to do this.
As we worked through this and started reaching out to vendors, we were experiencing a chicken-and-egg situation. We really hoped that we would have wide adoption and we would become the standard for plans, and we would have a lot of plans join us in this initiative. The plans on the other hand didn't really want to join until they had verification that there were going to be a lot of vendors using this information, and beyond that, a lot more providers using the vendor software.
On the other hand, the vendors were not interested in coming on board until we had a large number of plans. So we reached this impasse where it became clear we really needed to gain more experience to understand what it was going to take us to move to the next step.
Market testing of these processes was essential to developing a system that works for all parties, yet also can be adapted to the individual needs of the participating organizations.
In April of 2003, we launched a pilot with MedStar Health, which is the largest integrated delivery system in this area, with Doctor First, an e-prescribing vendor who I know you are going to meet with tomorrow, and with the local Safeway organization.
It was a year-long pilot. We were focusing on understanding the way the information would be used, could be used, and what it would require to get physicians to adopt and to use the technology. We had more than 100 providers in the MedStar system participating in this pilot. The year was just completed, and the final results will be out in the next several weeks. But I would like to share with you today some preliminary findings from this pilot.
The first thing is that drug interaction warnings were heeded four percent of the time, based on the information they got once they put the drug into the e-prescribing software. Likewise, approximately six percent of those prescriptions were modified when allergy warnings were made.
The drugs prescribed by the provider may change when the formulary exclusion date is provided at the point of care. In our pilot, when exclusions were given, 25 percent of the prescriptions were modified so that the drug is changed from non-formulary to formulary for the patient at the point of care, which of course means when they got to the pharmacy, they weren't surprised at the cost of the medication.
E-prescribing systems like Doctor First have a great potential for advancing the use of generics and for ultimately cutting the cost of the drugs prescribed. The pilot demonstrated that more than 75 percent of the brand name drugs prescribed could be substituted for generics. The final data that we have in the next several weeks will tell us how many of those were changed, but that 75 percent was a pretty startling number to us.
Some other findings that we had through our pilot. I mentioned the chicken and egg before. I think our estimate is that about 70 percent of the health plans are participating to impact the physician, to encourage them to use the e-prescribing software or the formulary, that is. Again, in a market where there is a dominant player, you probably don't need that kind of penetration of plans, but in a fragmented market such as this one, you need at least 70 percent of the plans to make it meaningful for the doctor.
There is also a need, we know, to ultimately link formularies to the patient level. At this point, we are providing the plan level formulary, which certainly is meaningful, but ultimately we really need to get to the patient level. It is certainly going to be costly, and there is a long term but very desirable goal if this is going to be successful.
Lastly, the adoption requires the allocation of significant financial and non-financial resources. Investments in IT and particularly in training and work flow adjustment are significant. At this point, much-needed clinical, financial and other incentives are being developed and piloted, and we support those forcefully as a way to advance this effort.
At this point, CPHQ is refocusing our efforts to promote formulary check as an integrated part of the e-prescribing system. The stand-alone website, I think we have learned, is really insufficient to meet the needs for most practices. It was a short term solution, and we learned an awful lot from it, particularly around the data standards that are required to provide the information in a meaningful way.
In March of this year, we began a partnership with RxHub, who I know you heard from this morning, to integrate our health plans' formularies to the formularies and PBMs that RxHub represents, into a more comprehensive centralized source of the data. The formulary information from all of our member plans is being combined with the information from all of the RxHub PBMs, and we have taken pieces from both of our standard formats and integrated them into what is now the RxHub standard format. We can now go out and represent to the vendors the majority of the commercially insured population in the country on this database, so we are not going to have that same issue with the vendors that we had before around the percentage of the market share available. This database is accurate and timely, it is quality checked, and it is complete.
We really believe that we have created the standard for the pharmacy benefit information for the industry, and as we move forward and better understand the way it will be used, we plan to enhance and adapt it.
We have had our learnings over the last three and a half to four years. I think that work that we have been doing has been important. As the rest of the e-prescribing industry moves along, we have become engaged in the e-health initiative. I'm sure you have seen that report. I think that the learnings together, all the stakeholder learnings, are pretty powerful to helping everybody understand what it is going to take to advance this technology.
The standardization we think is among the most important factors. The fact that this group is looking at that and going to make recommendations is very encouraging to us as we continue our work.
At this point, I'd be happy to answer any questions.
DR. COHN: Thank you very much. We will have discussions and questions after all the presenters. David, would you like to go next?
DR. MEDVEDEFF: Good afternoon. Again, my name is David Medvedeff. I am the Vice President for Government Business for Told Standard Multimedia.
Just a little bit of background. I am a pharmacist by training. I have been on the payor side for a little bit, also in retail pharmacy, and now I work at Gold Standard. Gold Standard as an organization is a company that for 11 years has focused on the development of a clinical drug database solely for the purposes of a clinical reference.
An example I have heard before is, we are the PDR on steroids. It is an electronic database that we license out into all different areas in the health care market. For example, eight of the ten largest retail chains are our customer, about a thousand hospitals.
What I want to do this afternoon for about 20 minutes is give you a high level overview of the Florida Medicaid program. We were the company awarded that contract. I'll give you some history of how it evolved, what the program looks like today, and then through the Q&A section, I believe I can answer more anecdotally on provider feedback and any questions you may have.
I like to begin with the end in mind. What we have seen in Florida is that we have had improvement both in utilization and on the clinical side. We just saw a statistic of about four percent of the significant drug interactions were caught in the pilot. Actually, our numbers are showing that there is about a four percent absolute reduction in significant clinical drug interactions in the physicians' offices that are using the PDA and the e-prescribing application.
Where we see an improvement in utilization is optimization of the pharmacy benefit. Florida Medicaid went into this as a way to impact the appropriateness of care, and also control costs. What we see is that duplications of therapy are cut to the minimum, and the fraud and abuse component which runs rampant in many Medicaid systems is also easily caught by this program and acted upon by the physician.
Why was Medicaid interested in this program, and why at this point in time? We have to look at what is going on in states and industry as a whole. You have to do that in comparison to managed care. When managed care has a fiscal crisis, in particular in the pharmacy benefit, there is a lot of cost shift that goes back to the patient. The new buzz word is consumer driven health care, where you are essentially given your debit card and you go out and you purchase your own health care.
On Medicaid, we don't have that opportunity, so what we have to do is develop some complexities in the pharmacy benefit that create what is known as hassle factors to help streamline the delivery of medications. So ultimately, what happens is, the provider is the one who has to get it through these hoops to get access to the medication that is appropriate for the patient.
So things that are happening in the Medicaid systems around the country, and particularly I would say at a greater pace in Florida, are things like preferred drug list, clinical PAs, step therapies, fair firsts, ran limits and prescription quantity limits per month.
So in Florida Medicaid, a patient can only get four brands in a month. At one point in time a few years ago, they could only get seven prescriptions in a month. It went from a completely open access formulary to a closed formulary. There were discussions of fail firsts. So what that means as a physician is, if I use drug A today, tomorrow I could require a prior authorization because it was a non-preferred drug. I could go through that hoop, and then three or four months later it might require prior authorization because now, a patient has to fail first on a different drug. It was very difficult for providers to catch up, and more importantly in the Medicaid world, it has a significant impact on provider recruitment and retention.
So what Medicaid did is, they looked at technology. At the same time that the Institute of Medicine had published their Quality Chasm report, Medicaid was looking at solutions to provider recruitment and retention, and they focused on technology as the lever there.
While all of this was going on in a changing benefit structure, there were also political pressures. We had newspapers in Florida that were publishing reports that five Floridians were dying every single day as a result of the misuse of prescription drugs. When they would follow the paper trial of where those medications were coming from, whether they were narcotics sold on the street or cancer therapies sold on the black market, Medicaid was paying a disproportionate number of prescriptions legitimately that would end up in illegitimate circles.
What those newspapers didn't focus on is that Florida was not alone in this statistic. If you look at some of the studies that are out there, about 200 patients die every single day because of the misuse of medications. I like to make an analogy there, as a Boeing falling out of the sky every day. That doesn't happen because of technology, because of air traffic control, because of all the computer systems. As the Institute pointed out, health care was lagging behind.
How the program evolved in Florida. In 2001, at that time Medicaid had what they called a voluntary preferred drug list. The concept was, Medicaid went behind closed doors with a PAT committee, they came out with a list of voluntary meds that they thought were the most appropriate to use first line. They had no way of communicating that voluntary list in any kind of real time fashion with their providers. So letters would go out twice a year with the voluntary list.
Around the same time, maybe six or eight months after -- the voluntary list never really got traction -- Medicaid looked at bringing a PDA in for the voluntary list, but that voluntary list turned into a mandatory preferred drug list. So now the hook in implementing technology and in particular a PDA in the physician's hands was, every percent that we can maximize our preferred drug list, we know we can save X millions of dollars in drug rebates and minimizing costs.
What happened very quickly is, the physicians adopted the preferred drug list. It was a very broad list. It was easy to use, and there really wasn't much left to squeeze out of the 95 percent compliance with any kind of formulary or preferred drug list. So we had to retool this. We took a focus on patient care, patient safety, and the positive side effect that was on the fraud and abuse impact.
In 2003, the RFP was signed, sealed and delivered, and Gold Standard won the contract, and we began implementing with Florida Medicaid. August 1 of last year was when our first doctor received their PDA. From that point going forward, everything we did working with the state was focusing on over prescribing. As we looked at some of the things being overlooked and the tools that are out there, there is this opportunity to impact over prescribing and under prescribing.
How does the application look at over prescribing and how do we define that with Florida Medicaid? First is appropriate drug information. The example I like to use is, when Prozac first came to market, a physician knew Prozac treated depression. The patient comes in, gets Prozac, calls the doctor a week later. The doctor at that point may or may not know that it takes three or four weeks to have the optimal effect of Prozac. If the doctor didn't know or the nurse intervened and didn't know, the dose was titrated early or another therapy was added, we have now over prescribed to that patient. We have driven costs, and about three or four weeks later we are starting to drive morbidity, when the patient has dose related adverse events or has drug-drug interactions as a result. So inappropriate drug information is where we start to look at over prescribing.
The other thing you can look at in appropriate drug information is treating lipids, are you treating the right lipid with the right medication, are you treating the right cause of hypertension with the right medication, it goes on and on, is it transferable between disease states.
The second component we look at is patient information. We heard today that this was the most important component of data that is being pushed today through electronic prescribing applications. We actually took a survey of our doctors, and this ranked the highest in value as to what the doctor uses the application for. It is for the patient specific drug history.
I know this is a meeting about standards and sets, and Medicaid is a little bit unique, in that it is a relatively closed system when you consider a patient who is on a medication benefit through the state does not usually go outside of that benefit to get medication. So 99 percent of drugs taken by a Medicaid patient is paid for by the state. So to get drug history, there is only one payor to go to, there is only one data feed to get, and we can present that to a doctor in a very quick fashion, and it can be done at the point of care.
Then the last component we focused on in over prescribing is fraud and abuse. That is the narcotic shopper, and I will show you an example of that. But fraud and abuse in Florida takes on a little bit different perspective, and this also goes into other states, and is transferable.
Last year, a grand jury convened to look at medication and fraud and abuse in particular in the Florida Medicaid system. The fraud and abuse issue with Medicaid was $1.5 billion, and $300 million of that is directly attributed to the drugs spent. So when you talk ROI and investment in the state's behalf, and the technology and electronic prescribing, it doesn't take a huge impact to have a positive return to the state, with a fraud and abuse problem that large.
There is a lot of information on the slide, but I will summarize it for you. The concept of looking at over prescribing, giving them good drug information, giving the doctor good patient history, giving the doctor a tool to impact fraud and abuse, was this a valid concept to take to market, and would doctors use it.
This was an e-mail that I received personally from the director of pain management at the Cleveland Clinic at South Florida the first week we announced the program. The interesting thing here is, we had only announced the program in Orlando, and this doctor was about 300 miles away, and sent me an e-mail that he wanted to be a part of this.
Basically, what the doctor is saying in this e-mail is that, my patients have pain, I want to treat them appropriately, but I don't know their drug histories, and they are poor historians. I have poly pharmacy issues, I have adverse drug event issues. This tool is going to revolutionize the way I do practice.
I mentioned the fraud and abuse piece. I don't mean to keep harping on that, but there is some interesting data that comes out of that. Part of the application that we developed with the state was the ability for the doctor not only in the future to talk with the pharmacy, but also to talk back with the payor directly about opportunities to improve utilization. So the doctor actually has what we call the DUR button on the PDA itself that messages back to Medicaid when the physician needs help, whether it is a train wreck, the patient is on 15 different medications, seeing five doctors, they need somebody to bring some continuity between that patient's care, or there is real fraud and abuse for narcotics.
Here is an example of one patient that was actually flagged through Medicaid. This is one of dozens that we get every week. What the doctor actually saw at the time was a 68 prescription history, because that is what we are presenting. The dates on this are five days, October 27 through October 31. This is what the doctor saw in this one patient's profile, ten different medications in five days, three highlighted medications are three atypical anti-psychotics which cost the state over $1,000 a month. You can maybe justify using two of these, if you were trying to work the doses and titrate doses to minimize side effects. What we had going on here, we had three different doctors in five days, you wrote a prescription for an atypical anti-psychotic.
The theory is that at the pharmacy level, that will be caught. But if the patient is going to three different pharmacies and each pharmacy is working in a vacuum, then the ultimate concept is that the fiscal agent or the PBM or the payor will catch that. Most of the time they do, but pharmacies do get bombarded with DUR messaging. When you try to make sense of the message in a vacuum, it is very difficult. Oftentimes that message gets overridden.
Here we had a sick patient with three atypicals in five days. The story gets better. The same patient had two antibiotics on two different days written by two different doctors. The patient also had two different NSEDs written by two different doctors in three different days. So this one patient had ten medications filled in five days, and seven of them duplicated in some form. It was costing the state about $1500 a month.
So when you talk about an ROI, it is very easy to see how presenting good information at the point of care -- and this is why doctors also use it, because the doctor wants to make the right decision, and they finally have the tool to do that in Florida.
Here is your typical doctor shopper. This was the full 60-day history. September 19 through November 19, and the narcotics are lined up there on the screen for you; there are probably about a dozen there. When I first saw this as a clinician, I saw methadone and I said, the patient is obviously an addict and being treated. What was happening at the time was, Medicaid was doing a switch, and they were doing drug A for drug B. Drug B was methadone. So this patient was not only getting methadone, but all of the other narcotics. He was probably selling them on the street, because we probably would not have seen him in November if he were taking all these.
I'll talk a minute or two about the program. As I said, we started August 1 of last year. We did a very low key marketing campaign, and we targeted doctors. What we heard today was, we need about 70 percent of the plans to participate in some type of application before there is traction and the doctor will use it. We looked at that a little bit backwards. We said, who are the doctors seeing 70 percent of the patients that are Medicaid? That is the low-hanging fruit, and that is where the state can see the return. So we focused on the highest prescribing doctors for Florida Medicaid.
We went into those offices and we offered the program to them. So we had about a thousand doctors that represented about 40 percent or 50 percent of the Medicaid pharmacy benefit utilization. What that means in dollars is about a billion dollars in just that thousand doctor cohort that we have, and that represents almost 14 to 15 million prescriptions a year.
So we could go in and with relative comfort hand a doctor a PDA. The doctor puts in a username and password, and the majority of their practice will be there, pushed down to them in real time, and they can see the Medicaid preferred drug list, all the plan design, the brand limits, the clinical prior authorizations, as well as the prescription history.
We actually had a thousand doctors enrolled in about three months. It took us about five to six months to go out and meet every doctor face to face, train them all face to face and support them. We have 500 doctors on the waiting list. What that means is, we still have doctors today signing up and saying I am interested. Because of HIPAA, we monitor the Medicaid server. If we see a doctor is not using the application, we contact the doctor to troubleshoot any technical problems. If the doctor fails to use it at that point, we take the device and give it to a doctor on the waiting list. The doctor who wants to use it and sees value in the application can start using that in their practice.
We have been approved in Florida to expand to 3,000 doctors. That represents 80 percent of Medicaid prescription volume. That 3,000 doctor number is not an accident. Governor Bush said he wanted to do the 80-20 rule in Florida, so 3,000 doctors within the next six months will be using the application. The bullet points on here state that this is completely turnkey. We go in, we train, we support, we provide everything A to Z for the state.
Where is all of this going? We took a very deliberate crawl-walk-run approach. When we first went out on August 1, we had a very dumbed down, very simplistic application. The doctor could only see the drug history. The doctor can only see what was on the preferred drug list and the clinical data. We didn't have much around the benefit design, and there was not electronic prescribing at the time.
What we then saw was, the doctors who were the early adopters started calling us saying, we want more. So we had about a thousand doctors in the focus group telling Gold Standard and Florida Medicaid what they wanted to see next, to the point where this completely evolved into an electronic prescribing application, which started rolling out a couple of months ago.
Everything that we have on the PDA today is also available on the website, so we allow the sharing of this information who cover for each other, as well as nurses who cover for doctors, et cetera. That can all be supported on the website, as well as on a PDA that the state has given to the provider.
I mentioned the expansion. We are also incorporating all the Medicaid HMO data which represents about a million lives in Florida. All of the largest HMOs in Florida that are providing Medicaid beneficiary coverage are now giving us their data, so that a doctor will not only see a fee for service patient, but an HMO patient as well on the device.
We also talked about labs at one point. The state has sent out an RFP for all the lab results. The lab results will begin feeding into the system as well.
With regard to standards and code sets, everything we do on the clinical side comes from our proprietary database. That is very transferable. We can talk with other databases if we need to, but at this point in time everything we do is on our proprietary clinical database. If you are First Data Bank or MediSpan, similar business rules.
From the transaction support side, it is driven on the medication list by NDC, and on the transmission it is the NCPDP script standard.
As far as where this goes from here, we want to embrace whatever the committee comes back with in the bill. We know that standards are necessary and we want to work with those, but we are willing to sit back and watch where the industry takes it.
Thank you for your time.
DR. COHN: Thank you. Robert Mandel, please.
DR. MANDEL: It is a pleasure to be able to participate in these conversations about the importance of standards and supporting the adoption of electronic prescribing technology.
In my written testimony, I highlighted what we believe at Blue Cross Blue Shield of Massachusetts are the key elements of information and functionality that are critical to the technologies, and as a result would benefit from having defined enabling standards.
We have worked with our vendor, Gixt, to answer the question specifically about technology. I am not a technology person by training, so they have helped us fill that out, and it is available in written form.
I think that our belief is, establishing standards will increase the likelihood that any system a physician adopts will provide the functionality that will be most beneficial to the prescriber, the patient and the system as a whole. Furthermore, it will hopefully rationalize any efforts there are in markets where there are multiple entities trying to develop the standards.
I wanted to focus this afternoon specifically on our experiences in e-prescribing, both in a pilot that we have done in Massachusetts, as well as a collaborative program that we have put together with one of the other health plans, and talk about some of the hurdles that we have identified, what we have learned about implementation, and how this might benefit the committee as they look at standards.
We have recently had some focus groups of both physician adopters and non-adopters who have been involved in some way with the program, so I would also like to provide some information from that.
Just to describe the program briefly, our pilot program was implemented with Pocket Script, which has subsequently been bought by Gixt. It was offered to physicians based on their expressed interest in electronic prescribing. Access was provided to all the prescribers in the practice of physicians, nurse practitioners, and the application was provided through a hand-held device that communicated with a local server in the office via YFI Technology.
High speed Internet access allowed the server to retrieve real time eligibility, patient specific formulary information as well as medication history for the patient. So those were all available to the prescribers on their hand-held at the point of care. The program was provided at no cost to the providers, and obviously training was provided as well as ongoing technical support by the vendor.
MR. BLAIR: Did you say patient history or medication history?
DR. MANDEL: I'm sorry, medication history for that particular patient.
The technology provided through the collaborative program is a bit different. The data is now hosted remotely on servers at a Gixt site, and communication is through either access points and Internet access in the office, or through cell phone coverage. So you can have different types of hand-held devices; some are cellular based and some are wide area network.
The demographic data in both cases is preloaded, so we typically will take a feed from the physician's practice management system and load all of their current or existing patient demographics, so that they are available in the system when the doctor sees his patients.
That doesn't help new patients. Those new patients are required to be loaded in both their practice management system and typically the e-prescribing system. The e-prescribing system requires about five data elements, so it is a pretty brief process. For primary care physicians who typically don't see a huge number of new patients a day, it seems to be fine. There is a little bit more difficulty with specialists, and we have developed some interfaces so that there is direct input, and they only have to enter it into their practice management system.
We saw initially a lot of interest in the pilot and we have subsequently in the program. What we have learned is that interest certainly doesn't translate into adoption. There are a couple of reasons for that. I think physicians in general -- a number of physicians are interested in new technologies, obviously are interested in ways that might improve the efficiency in their offices. I think electronic prescribing is potentially offering that to them initially. However, as soon as they encounter one difficulty in using the technology, it is very easy to put that down and pick up their prescription pad again and move forward.
The benefits that they are going to see, which are typically for their practice, reduce phone calls and interactions with pharmacies, and potentially with patients calling in, because at the pharmacy they have had a higher copay than they wanted, don't tend to be experienced for several weeks at least, and potentially even longer than that. So a problem early on without having seen the benefit yet, it is pretty easy for the physician to just put down the device and walk away.
Subsequently with the program, we added in addition to providing the program at no cost at least initially, we added incentives to a primary care physician incentive program we have, so that physicians who implemented electronic prescribing would receive a dollar per member per month of payment if they successfully implemented and continued to prescribe at least 50 percent of their prescriptions electronically. And of course, that is based on Blue Cross Blue Shield members, but we are assuming that they don't just pick it up for our members.
That incentive has just been put in place. We have had a huge amount of additional interest as a result. We have a waiting list of over 1800 physicians that are waiting to get the devices. We had more of a low touch approach initially, where physicians could register on line, they could call in, the devices could be shipped to them, they could get training over the phone, they could get training in forums or either directly in the office. Training would last an hour, hour and a half.
What we have learned is that again, without on site support with that first couple of problems they encounter to help them get over that hurdle, they are very likely to walk away from it. In the focus groups we recently did, among the adopters it was amazing to hear the functionality they didn't know was available through the application. They had had training, they said the training was excellent if you asked them what they thought of the training, but a lot of their complaints about the system were for things like, we would really like it to provide medication history. It does provide medication history. They obviously didn't remember that or didn't know how to use it.
So our takeaway is that we need a lot more hand holding initially as the physicians begin to use it, to get them over that hurdle, to make them feel comfortable. Hopefully after they have used it for a couple of weeks, they will feel comfortable. If they run into a problem, they will be willing to call and see enough value in it to go back to it once they have gotten that problem resolved.
The other things that the focus groups talked about -- and this was both adopters and non-adopters -- the cost of the technology is an issue for them. They don't feel that this is something that they can afford to do, given the constraints of their practice, increasing costs, particularly malpractice expenses, the limited increase in revenue based on the rate of increase that insurers and Medicare are increasing their reimbursement, creates a lot of tension for them in terms of what kind of technology they want to invest in.
Again, it was interesting that although they recognized that there are significant issues nationally with medication errors, and they understand the reports from the Institute of Medicine around medication errors and deaths related to that, they aren't the ones making the mistakes. So they don't feel the need on a safety basis to invest in this technology. They are looking for a return on their investment in terms of efficiency and productivity in their office. Again, they recognize there is a problem with medication errors, but it is not their particular problem.
Then to expand that in terms of what other kinds of benefits could derive from the technology in terms of safety and quality, we posed a question. This didn't come up in the group on its own, but subsequently when the moderator comes back and asks if there are any questions from the observers, we posed a question about compliance, whether they had any thoughts about compliance.
Their initial thought was whether we were talking about their compliance. Then one of the doctors said, I guess we could potentially see that someone didn't refill their medication, and if they were having symptoms that we needed to do something about it. But they didn't really think about this as a way of monitoring compliance initially. I think I have seen three newspaper articles in the last month about the fact that 50 percent of patients don't fill their prescriptions. You would think in terms of quality of care that that would be a big concern for physicians, but they have a huge level of frustration, given all of the time constrains and everything else in their practices. So the thought of taking on an additional burden -- and I think that is how they viewed this -- of having to monitor their patients' compliance or be concerned about it was something that they weren't thrilled about.
So I think we need to think about ways of making the technology something that they can endorse and seek benefits from through either incentives or other things that allowed them a little bit more freedom in terms of the economic pressures they feel to approach the technology.
Finally, I think there is a very low threshold for frustration. I can't tell you how many of them said, I tried it, I ran into a problem, or I used it on the days when I have a really slow day, but on the days that I am really busy, I use my prescription pad because I think I am faster with that.
Again, they have a system in place that handles all of those calls from pharmacies that are unnecessary because they didn't write a medication that is on the formulary, or they wrote a third tier medication instead of a first tier medication, and the patient has a $35 copay and is really unhappy about it. Their office staff takes care of that, they are staffed for that. They don't really see that as -- that the potential to eliminate that would be something that benefits them. It is, I'm comfortable doing what I'm doing, it works the way it is now. Give me a really good reason to do something different that I have to work at that may cause a little bit of pain and suffering for me.
So I think that is a long way of getting to the point. I think the creation of standards that allows physicians to feel comfortable -- and I am not necessarily talking about mandatory standards, but standards that would encourage all of the vendors to create similar functionality and similar ability to communicate among health plans, and whether that is doing prior authorization electronically or whether that is communicating with the pharmacies, incorporating medication histories with the pharmacies as well as the pharmacy benefit managers and other sources, allowing physicians to be relatively secure in the fact that that is going to be available when they select a vendor, will make this a much less fearful event.
I think right now, they are concerned that if they pick one vendor, that vendor may not have all of the information they need, and they are not really the kind of people that can decide. We get a lot of calls at Blue Cross saying, can you recommend a vendor on electronic prescribing, can you recommend a vendor for an electronic medical record, this isn't really what we do, and we have looked at 15 vendors and we don't know which one to choose. So I think having some assurance that whichever one they choose, they are going to get the basic functionality they need would be very reassuring for them.
That is a benefit from the ultimate outcome, in terms of the benefit it delivers to patients, to physicians and to society as a whole, but also as a psychological benefit of the physicians feeling more comfortable in their decision.
I think one of the driving forces behind our collaboratives, and it is a bit unique for two competing health plans to collaborate on something like this, but we both have done pilots with the same vendor and had very positive responses from a significant portion of the utilizers. Again, the significant portion started and didn't continue. Again, that is a part of why we did the focus groups, and why we are modifying the way we implement, to address that. But we felt that it would reduce confusion in the marketplace if we approached it together with a single vendor initially, so we wouldn't have physicians wondering, should I choose this vendor, should I choose that vendor. Blue Cross is offering me this program, and 80 percent of our network overlaps, so we would be -- and probably our hard subscribers are even higher overlapping -- we would be targeting all of the same physicians, sending out mailings. They would be trying to decide which they should choose, and given that dilemma, may decide not to do anything or wait and see what the next offer is going to be. So we wanted to try and avoid that.
I think we have been commended, and I think the physicians have been very receptive to having a combined program like that. I mentioned that we have incorporated in our primary care incentive program an incentive for physicians to use an electronic prescribing technology. They can also receive the incentive if they have an electronic medical record or health record, or if they have an electronic decision support functionality. So it is not just e-prescribing, but e-prescribing is one of the things that is easier for them to adopt.
Our intent is, and we have done this with our incentive program over the years, to continue to raise the bar in terms of what is required to receive the incentives. So we start out with electronic prescribing at a certain level, and then move to a higher level of electronic prescribing or a more robust system with more decision support capability or more history capability.
Certainly based on the feedback we have from physicians, providing the technology at no cost is necessary, but not sufficient. They really need a lot of support around adoption and recognition of the changes they have to make in their office to do that.
I think that the initial evidence we have from the pilot suggests that there is savings, and other people have shown that there are significant savings in a commercial population as well as the Medicaid population related to improved generic prescribing. Just to give you a scope, a shift of one percent in generic prescribing for our plan represents a $10 million savings. So clearly, if we can shift prescribing and increase generic prescribing through therapeutic substitution, through messaging on the devices, there is money to provide for rolling this kind of technology out and for payors ultimately to support this kind of technology.
I guess the take-home message would be, anything we can do as plans or government to support adoption through making it easier for physicians, making sure there is a single set of standards there that make it easier for them to understand that choosing a vendor is not going to impact how they can communicate or who they can communicate with, is very important and will go a long way to making sure that this is successful.
Thank you.
DR. COHN: Thank you very much. Laura Adams.
DR. ADAMS: Like Robert, I don't have Power Point or slides, and that is probably okay with the group. I think by this time today you are probably a little worried about Power Point poisoning anyway.
I would like to give you a little bit of background about the Quality Institute in Rhode Island, so the information is germane to how we are moving forward with implementation of electronic prescribing there. We are a collaboration of hospitals, health care providers, organized medicine, organized nursing, pharmacists, insurers, business, academe and government. We are organized for the express purpose of improving health care quality, safety and efficiency in Rhode Island by an order of magnitude.
We have five principles that are probably relevant for this discussion. The principles that govern our behavior as an organization are first and foremost collaboration. Because we have people who are, some would describe as natural enemies, it is probably not too much of a stretch to say that it is hard to breathe the same air sometimes in a room together. We did set aside a principle from the very beginning that said we will behave as collaborators. If that doesn't sit well with you, now would be a good time to stand up and mention that.
Then we also said real improvement is required. We are not a think tank. We are not here to produce yet another white paper. Our job is to move the needle in the state of Rhode Island. We are also looking for win-win solutions. We do not believe that we can sustain ourselves as an organization if factions can pull together and attack other factions.
We also indicated that we are here to focus on system improvements that none of us can achieve alone. We believe that by now, our problems are bigger than any one of us. The idea of continuing to move forward independently, that time is past.
Our last principle is probably the most important principle. That is, top leadership, please don't send your lieutenants. We said from the very beginning that we want the CEO's, the highest ranking people in the organization around the table. We don't want five months from now your assistant director of landscaping coming in your place. We have seen that happen too many times in organizations, and we have a big job ahead of us if we are really going to design this health care system.
We also set out for three areas of strategic focus at the Institute. They are patient safety, improving the health care information infrastructure, and evidence based medicine. It did strike me one day -- I have to admit, I did drive through the drive-through at Wendy's, and watch my order come up on the screen, and realize there was more technology in getting my hamburger than there was getting my cardiac medications to my patients. We are behind, and it is indefensible.
So we took this on as our first project, electronic prescribing. The purpose of that was, we knew that it was compatible with the Quality Institute's principles. That is real improvement. We believed it would be a win-win. We believed it would be politically palatable. We wouldn't be destroying any major force in the state by trying to do that. And it would require the cooperation of a great deal of us in Rhode Island to make it happen.
It did advance our areas of strategic focus, specifically patient safety and health care information infrastructure. Over time we will advance evidence based medicine through the same pipeline. It gave us the opportunity to gain some skills that would equip us for larger scale project work, which we found to be very important, because as you will see as I move later on into the discussion, we are somewhat convinced that going application by application will not be sustainable in the future. We are finding that to be a great difficulty right now, writing an interface for every new piece of technology or every new application that comes down. That has been a formula barrier for us in Rhode Island.
We chose to work with SureScripts in our electronic prescribing initiative, for several reasons. First and foremost, it was consistent with our principles. These were the large chain drugstores and the independents all agreeing to work together, so it was competitors collaborating, and that struck a chord with us.
They are not a software vendor in that case, and they are also the clearinghouse for us. We have an open platform agreement. SureScripts' philosophy is the idea that they are not there to influence, to push a physician to prescribe in one direction or the other necessarily. It has linked any certified retail pharmacy to any particular electronic prescribing vendor that chooses to get certified, so we like that idea of that completely open platform.
We knew once we installed this kind of plumbing, however bare bones and basic it was in the beginning, we could build on it tremendously. But the most important aspect that governed our selection of SureScripts as the direction for us to go is, we knew that electronic prescribing in Rhode Island as it was defined at the time was not electronic prescribing. It is, hit the button, produce a fax at the pharmacy. The fax is then hand transcribed into the computer system at the pharmacy. What we saw was an incredible lost opportunity in Rhode Island at least for the ability to get refill information automated.
We kind of thought that might be important in the beginning. We had no idea how important that was. That is the value proposition for our adoption strategy in Rhode Island, and it works. We have seen significant efficiency gains from that. So we believe that without that complete electronic pipeline -- and we know that there were places in the nation where one or two particular drugstores made an automated system into their physicians' offices, but this was our first opportunity that we saw to make that ubiquitous in Rhode Island. So the key here is the efficiencies that are gained in automated renewals. That is available at least for us in Rhode Island only through SureScripts.
The statewide electronic prescribing project background. We really are in some ways an ideal learning lab for the nation. We are very small in Rhode Island, no bigger than the city of Minneapolis in terms of population. You can get everybody in the state together in 15 minutes for a meeting, so it makes it quite conducive to this kind of work.
We think that SureScripts chose us because it was one of the states in the nation, one of the few states in the nation, where they could sit down at one meeting and have every significant stakeholder needed to make the system work available in one meeting, and with a commitment to collaborate and now, for us, a history of collaboration.
We have 2,000 active prescribers in Rhode Island, not many, but 2,000. We constituted a professional advisory team in January of 2003. This team is comprised of key opinion leaders and early adopters. We didn't choose high prescribers. We chose purposely those that could influence others amongst respected clinicians, those that we knew that if it made sense to them, it would make sense to others.
We chose our leading pharmacists, educators, University of Rhode Island College of Pharmacy, researchers and representatives from SureScripts and the Quality Institute formed this professional advisory committee. It was absolutely essential to getting adoption going. They were able to shape how the adoption happened, shape some of the early rollout strategies, and in some cases even shape how we applied the application.
The professional input team interacted closely with the Rhode Island board of pharmacy and the Rhode Island Department of Health. I have to say that when I began to understand the power of the collaboration, I am most energetic about this. I wish I didn't have to sleep so I could work 24 hours a day in moving this thing forward. But sometimes I get a little ahead of myself. I was out talking this up in Rhode Island, and someone leaned over and whispered to me as I was mentioning that we were getting ready to do the beta, and they said, do you know that it is still illegal to electronically prescribe in the state of Rhode Island? I thought, that is embarrassing, no. So I went with hat in hand to the Department of Health director and mentioned this little point to her and she said, Laura, I'm on board, I am on your team. This is my project, too. I have been putting all of the parallel work forward. As you are doing your work, I am doing mine. By the time we are ready to beta, it will be legal in the state of Rhode Island to do so. At that point, I realized that I didn't have to go to her later, try to convince her to get involved in our project and to do so post haste. She had already done the work because she was part of the team.
We began beta testing June 9, 2003 with 40 subscribers, and only 50 percent of our pharmacies linked. Still at the time, 50 percent was a goodly number of them. Our beta test was very successful. The technology worked. We ran about 10,000 new subscriptions transmitted through the system in the six weeks beta testing time frame.
I think it succeeded surprisingly well, despite what we now look back on as considerable problems with our early rollout, that stemmed primarily from it being the first footprints in the snow, the first time we were running electronic prescribings through many of our pharmacies. For some of our larger pharmacies, that is about .5 percent of their business, was coming through electronically at the time. So to have it succeed to the degree that it did -- .
We pulled together a focus group of beta participants that was completed October 2, 2003. The key findings there of those first six weeks participants were that the value of the renewal function is unquestionable. I will state that many times during my testimony today, to just underscore how important this is for an adoption component. It simplifies and streamlines a cumbersome process for them. It resulted, in their words, in a huge efficiency gain. Also resulted in improved quality from their perception, reduced opportunity for error, resulted in improved phone access for patients, and more time available for clinically relevant staff activities.
The value of the new prescription function was mixed for our early six week prescribers. They had particularly problematic during the adoption phase, because most of our pharmacies were early in their training. So we had a number of incidents where that electronic prescription would sit in the pharmacy unnoticed. The patient would get there and they would be told that there is no prescription for you. After multiple phone calls back and forth, they would realize that yes, indeed, there was a prescription there, but it was hidden in the electronic system. So our major retail drugstores in Rhode Island have undergone a tremendous amount of training of their staff and redesign of their system, so that electronic prescriptions, the alerts appear very clearly.
We had other problems, such as, their process was so fax based back to the physicians' offices, that even when a prescription was transmitted electronically, a renewal would go back as a fax. And because our prescribers liked this system so much, they found that extremely irritating, to be electronic, but yet receive their renewal fax notification. So we had to build in systems where if the prescription was received electronically, the only way to send the renewal back was electronic, so that it stopped the inadvertent faxes going back and forth.
Those are just a couple of examples of some of the things that we ran into, just purely as a matter of early implementation issues.
Also, the issue of physician work flow and portability of equipment. For many people, they had but one computer, so to go to the computer to do this -- I can't tell you how much we agree with Robert. In our experience, it has been the same. SureScripts and others were right there by the sides of our prescribers as they were trying this out in the beta test period, and it was absolutely essential. There were so many little snags, that they could turn to someone right there and get the information on how to get through that, how to learn about the system. We found that to be very, very useful, essential.
We thought that would go away after we got all the snags, and everything began to line out a little bit. Our pharmacies were much more on board. We don't think that is the case. We agree with Robert, that that will be an ongoing aspect of this, no mater how many physicians we get on board, will be the in-office redesign support and hand-holding that it will take.
We did find that our focus groups also felt that the application was good with specific enhancements and additional choices were desired. They of course want some of the same things that we see in the formulary. We don't presently have at this moment capability for formulary information on there, PDR type of information, complete medication records.
Our biggest struggle here with some of our more advanced physicians were the fact that they already had existing medical records, sometimes homegrown medical records. It was bloody hard to write interfaces to every one of those, and in fact, we didn't do it. We wrote interfaces to some of the larger practices, but in Rhode Island there are very few large practices. The vast majority of our physicians practice in onesie and twosie type situations in their office, not affiliated with a large group.
The issue of having -- we found that the less technologically savvy the office was, the easier it was for us to apply this one application and make it work. For more sophisticated providers, this did not interface, and this caused more work for them. They saw productivity hits they never anticipated from having parallel systems, one foot on one surfboard and another foot on another surfboard. So the use of the parallel systems has been a big barrier for our more sophisticated physicians.
Their expectation for a high level of personal on-site support was also stated clearly in our focus group information. They wanted more complete pharmacy participation. By the time they were able to send it to 50 percent, they wanted 100 percent. If we have had any kind of pressure coming from our physicians, it has been, get all the pharmacies on board, because this is so efficient for our renewals. Once again, it also creates a parallel system when they can send it to this pharmacy electronically, but not that pharmacy.
We ran into a slight problem in the beginning. We had to work with one of our practices there, a practice manager who is a very wonderful, bright woman, who is also a nurse practitioner. She said, I had to hold back every minute from telling my patient, don't go to this pharmacy, go to this one, because they do electronic prescribing. For us, that would have been the kiss of death because we were busy trying to roll this out to as many of our pharmacies as possible, and that had to be a stepwise rollout, and we did not want to harm some of our pharmacies for being later to the game than others. So we had to work hard to make sure that didn't happen, because the inefficiencies are just there.
Moving to our current status, we now have 70 percent of Rhode Island pharmacies connected and activated on the network. These include CDS, Brooks, Walgreen, Stop N Shop, Walmart and now QS-1, which serves a third of all of our independents in Rhode Island. So we have now just two-thirds of our independents left to bring on board.
We have 250 physicians activated now, but we have 1100 coming up in the summer, so it would be a big jump for us with this second tier.
We do have an impact study that is in progress right now. It is being conducted by the Abacus Group in Rhode Island. That will be surveying the effect on efficiency as well as satisfaction in prescribers, staff and consumers.
LifeSpan, who is our largest health system in Rhode Island, will be rolling out their system. That is part of our 1100 prescribers. When we are finished with the LifeSpan rollout, 45 percent of prescribers in Rhode Island will have access to the electronic prescribing system through SureScripts. That is approximately a year out from beta testing. It is going to be about 11 months out from actual end of beta test evaluation.
By the end of the summer we will have 80 percent of Rhode Island pharmacies, and our remaining 20 percent will be independent pharmacies that are not aggregated through a single system that we can reach them, for example, like bringing up QS-1.
Our critical enablers to promoting adoption in Rhode Island. We started with the true value proposition; word spread. The other thing that we realized early on is, get the staff involved, because that is who benefits, the people who are struggling hours and hours and hours to deal with the refill process. So we got them involved early on. They are tremendous marketers of this.
We insured that major stakeholders were behind this from the very beginning. LifeSpan CEO George Veccione sits on our board, and in fact is the chairman of our board, and we know that had a lot to do with LifeSpan coming on board as a major player and bringing 800 of their physicians on board in one fell swoop with this. It started out as a coalition; we did not have to convince them it was the right thing to do. They designed the project themselves in cooperation with the rest of us.
We had broad participation in design and implementation. We built in flexibility. For example, they can start with renewals only. They don't have to start with writing the new prescription to get the system up and going. In some ways, that has been a bit of a detriment, because again, we have the same experience Robert has, that if something goes wrong and there is a glitch or two, it is so much easier to reach over and grab the prescription pad and write it. So we are not seeing as many new prescriptions going through the system as we would like. We certainly aren't seeing as many as we would like, and we think that is part of the issue.
We also think that the heavy support again was part of the critical enabler.
Our challenges that we face now. The early work flow design, where our pharmacies were unaware of scrips, really did hurt. That was picked up and noticed by people. There is still a lingering fear that that is going to happen. On occasion, we still have that happen, where a prescription lands somewhere electronically, and it is not known that it is there. Every one of those of course gets so much more play than the thousands of other prescriptions that go through without a problem.
We had database inaccuracies that caused physicians not to be recognized on the system, because there was a slight difference in the way they were registered through our SureScripts system, and perhaps the way the pharmacy had them registered. So we are doing a lot to clean up the databases and make sure that they match.
Some of the more sophisticated and efficient physicians' offices had a real problem with work flow integration issues, and those remain. Not all of our pharmacies are on the system, so parallel processes remain in place.
The benefits vary. We have the experience of having the only vendor available to us at the time we were ready to beta test the system. The only one certified by SureScripts at that time was Oncall Data, System DX. So they got a big leg up in terms of market share, so we have a few market issues. Because they are now also the vendor to LifeSpan, when LifeSpan comes up we will have 45 percent of the active prescribers in Rhode Island, that is all Oncall Data. So we do now have a Doctor First certified on the system, but for them to come in and try to make a major push in Rhode Island for market share -- remember, no bigger than Minneapolis -- we are not seeing a lot of effort being spent, understandably. So it is more of the legwork that we will have to do as a coalition to make that happen. Of course, schedule two drugs are something that step out of the electronic system, and once again, it creates a separate parallel system for prescribing.
With our larger groups now signed on, we just have the offices of one to two and smaller groups now. There is not that much competition.
Just to summarize a couple more things, the desired features that are high on our list is real-time formulary information, which we understand will be available very soon. They would like to have drug prices available, that information, so they can instantly communicate to patients and have a conversation about the drug prices. Community-wide drug databases.
They do want the notification feature, whether or not a patient filled a prescription. However, as Robert mentioned, we have had pushback on that. Some have said, my God, that would just mean that we would have to go out and follow up on this, and there are even greater concerns, because if we do not, it will present us with a malpractice issue.
MR. BLAIR: I'm sorry, could you repeat this issue? I missed the beginning.
DR. ADAMS: Yes. The concern from our group about being notified that a patient did not pick up a prescription, their concern was twofold. One, that it would increase the work load of following up on that. Secondly, that if they didn't follow up, it would present a medical malpractice issue, that they knew that the patient did not get the medication and they did nothing.
However, we have had interest from a major medical malpractice insurer in Rhode Island, that if we were able to generate an automatic notification to the patient, that we noticed that you didn't pick up your prescription, it is important to do that, and simply sent the letter out as followup, and incorporated a couple of other, similar features, that they would be interested in joining with us and offering discounts to physicians for coming on to the system. We do think in this case it takes a village to convert physicians, so any little help that we can get, we are all for it.
So our benefits were that before our studies were available, we were operating off of anecdotal evidence from our year's experience now, almost a year. Anecdotal evidence from our practices and our pharmacies are that they believe that system makes care much safer and it is quick and easy to use, no reason not to.
The allergy information, the drug-drug interactions, the patient's name and demographic information at the top of every screen and the presence of dosing information improves the safety of the process. Anecdotal evidence tells us that when the system worked as designed, the patients love it. The reason they love it is because it prevents one more stop at the pharmacy. They don't have to drop off the prescription, go kill some time, go back and pick it up. They know that they can make one stop to the pharmacy.
In Rhode Island, we had a standard for most of our pharmacies that this is treated as a phone-in prescription if it is electronic, so there is a one-hour guarantee of availability. So we can tell the patient, wait an hour, the prescription will be ready. They can go home, they can run some errands, but it cuts out one more stop.
We have other anecdotal evidence that tells us that they feel much safer when the system is electronic. There is something about that idea of moving into legible prescriptions that is appealing to them. The irony of the Wendy's and the electronic has not escaped them as well.
Anecdotal evidence tells us that there is a significant efficiency gain. One of our ten prescriber offices found that they used 1.1 fewer RN FTEs dealing with medication renewal process as a result of streamlining it. This office gets 50 to 80 fewer phone calls a day as a result of being able to send that prescription in. They no longer have their patients call their office for details. They have them call the pharmacy, because it is the most streamlined methodology for doing it and gets the patient their refill the fastest.
We think over time that as these refills get -- as we get more sophisticated with this, we will be able to tell the physician's office well in advance, this is your next month's upcoming renewals, rather than the immediate for today, and they can begin to prepare and arrange their work load without having, here is your renewals today.
We have a documentation interaction. In pharmacies it is all in one place. Physicians handle the majority now of renewals directly. I would like to say that is how it was being handled before. We have some evidence that it was not. They were not as directly involved in renewals as we would like, but they are now. I think under our lessons learned, to close this out, we think that collaboration works and is essential. For us to move this forward in Rhode Island, it has taken lots of us working together to do it. We believe strongly in the wisdom of coalitions.
We think that when integration problems were inevitable, we knew they would be, but when those problems popped up, we first of all agreed to acknowledge there is a problem, and stop denying it, stop defending our own particular component of the process, but acknowledge that it is. Avoid becoming defensive about the source of the problem, be committed to finding solutions, and stick with it.
We don't consider barriers impenetrable. We consider them something to work through, so we stay with it. We avoid jumping to quick solutions. We do not install work-arounds. We find that we have to go through the discipline of performing a standard analysis when something fails. We do not put a Band-aid on it, we stop and fix it. That is another reason why we haven't advanced as quickly as we would like, but we feel very confident that we have a safe and reliable system here.
Training has to be done right prior to implementation. The knowledge retention drops off exponentially between the time they have been trained and the time they actually use the system, both pharmacists and physicians as prescribers.
Our last learning is perhaps the most profound for us. We will continue to move the process forward in Rhode Island and continue to try to get electronic prescribing adopted in Rhode Island. But we are now moving to the concept that this has to be part of a larger system. We have got to turn to bringing forward EMRs in Rhode Island. We believe Blackford Middleton's research that shows the pie diagram with the 11 percent return on investment when a physician's office implements electronic technology.
In Rhode Island, more than 90 percent of that return on investment goes somewhere else. So we have stopped asking the question, why don't physicians adopt. They don't adopt because there is no business case for it. It is very clear that if they practice extremely good evidence based medicine, and install IT systems, that they will be taking themselves one step toward insolvency.
So what we have decided to do in Rhode Island is understand that there is the other 90 percent of us that are gaining from their adoption of information technology. We have to step up to the plate, ante up, and pay for it, and we think pay for it in a big way.
Our next project will be to embed our electronic prescribing system in an electronic medical record, along with decision support, all types of decision support that are reliable and we can have confidence in in the beginning. We have to commit them to each other. It does no good to have stand-alone electronic medical records if you cannot get information on this patient from the other sources that contain information.
So we are going to install a demonstration project in Rhode Island. Our plan is to install for about 200 physicians in a community of about 100,000, an electronic medical record, electronic prescribing, decision support and connectivity to the entire community. That is to pharmacies, to the laboratories, to the imaging centers, to the hospitals and to specialists, each other, state agencies, mental health and long term care.
We have to pay them to use it. It will probably be our plan right now that it is $25,000 a year per year per physician to use it, and then as we -- we strongly believe the studies that tell us that there is approximately 30 percent waste to be had here, if we learn how to streamline these systems, to stop the duplication, to get the gains that we get from the use of evidence based medicine.
We think that our chalcedonies tell us that if we can capture even 11 percent of the waste, that is, the expenditures that are out there that go to wasted money, we can get approximately a 77 percent return on that investment. Even if we fall on our face and we capture only four percent of the dollars that are categorized as waste, there is still about a 20 fold percent return on that investment. We think it is worthy of at least a demonstration project and a trail. We may fail.
Some have said, that is high risk. We couldn't disagree more. We think that nothing is more high risk than standing still and doing nothing.
So that is where we plan to head in Rhode Island. We think it will be a big pull up hill, where we have to get all of our payors to agree to do this, all of our purchasers to agree to do that. There will have to be returns that go directly back to the purchasers, potentially not to the payors, but directly back to the purchasers. We have a big road ahead of us, but we think it is worthy of many years of our life doing it.
Thank you very much.
DR. COHN: Thank you all for a set of fascinating presentations. I guess I should ask the subcommittee, questions, comments?
MR. REYNOLDS: I have one. One of the presenters mentioned that maybe 50 percent of the patients don't fill the prescriptions. As we move to electronic, is that going to -- in other words, if it goes to the pharmacy electronically and the person doesn't show, what kind of an issue does that create as we move forward?
DR. MANDEL: We're not sure what kind of issue it creates for the pharmacist. In other words, can they put it in the bottle and then take it back out. But the point I was trying to make is, in terms of appropriate treatment and effective treatment, if patients aren't picking up their medications and aren't taking them, then they aren't being treated appropriately. I think that should be the major concern for the physicians, because they are responsible for the treatment. So either they are not educating the patient in a way that conveys to the patient the importance of taking the medicine, or something is getting lost in their interaction. But basically, that is wasting the system. They have had an encounter with a physician, they have had recommended treatment, and they are not doing it. How can they expect to get better? Or else, it tells us that 50 percent of the treatment isn't necessary. So either one of those is obviously a significant waste in the system.
DR. MEDVEDEFF: If I could speak to that also, just as a pharmacist, I was with Eckerd Corporation for many years, there is waste, because when the prescription gets to the pharmacy, the pharmacy spends resources in having that prescription filled, and then spends resources in reversing that prescription and putting it back on the shelf. So once the prescription is filled, retail pharmacies become very savvy at pulling that patient into the pharmacy through automated phone-out systems, through tickle files that they have installed in retail pharmacies.
So you have a physician who has spent time treating the patient, sent the recommendation electronically to the pharmacy. The pharmacy has now filled it, and now you have a pull to get the patient in the pharmacy to receive it. So you may see an upward tick in prescription volume, but we are also squeezing the waste out of the system, which will probably offset it. We are seeing the same amount of dollars, but we are doing it the right way.
DR. ADAMS: I think for our physicians, if they can work out those issues, their concern too is that patient shows up again with high blood pressure still, and they up the dose, and the add another medication. The truth of the matter is, the patient never took the first medication. So this is for us a consideration of overuse and safety issues.
DR. COHN: Other questions? Robin, I have a couple of questions for you. I was fascinated by a couple of your things. I wanted to find out a little more.
One of them has to do about this crosswalk being developed around therapeutic classifications. I was just curious if you could tell me a little bit more about the resources you have had to invest in that. Do you see this as something you will be maintaining off into the future and what its value may be to others?
I also wanted to ask you a little bit more about your de facto standard formulary database. But can you answer the first one?
MS. THOMASHAUER: Actually, yes. I am going to refer you to Gwen Lowe, who is our project director and who can give you a lot more detail than I can on that.
DR. COHN: Can you introduce yourself?
MS. LOWE: Basically, the plans --
MS. THOMASHAUER: Gwen Lowe is the project director for the formulary database project at CAQH.
MS. LOWE: I'm going to try to make this quick. Basically, the plans use primarily MediSpan and First Data Bank, the CAQH plans. We can't actually speak for the other plans, because we do not look at their standards.
But in order to build the website, which is actually no longer up, as Robin explained -- I don't know if everyone understood that.
DR. COHN: No, actually, I didn't.
MS. LOWE: We found that the website was only getting 7,000 hits per month. So basically, we wanted to focus on the vendor distribution, so we partnered with RxHub.
So when building the website, we needed the therapeutic classification system, so you had to look at MediSpan and First Data Bank, and those were -- the two systems classified drugs differently, so we had to take the two systems, make a comparison. When I say we, not the CAQH staff; we placed the pharmacy directors of the health plans. They sat in a room for five days, I don't even remember how long, it was a very, very long process. They came together and they said, we want our drugs to be classified directly. We are not going to stop using the systems we are using, but in order to put them on the website and be properly loaded, you need to build a crosswalk between these two companies.
So the resources that were put in were the time of the pharmacy directors of the plans, which is significant, bringing them together. They had to fly -- there was one in-person meeting, months of discussion, then also building the website. That was operating at almost a half million dollars a year to keep that up and running, plus another half million to build it. So a million and two, the website, all the time and effort to build the classification crosswalk, and then the decision was made that rather than have this one stand-alone piece on the website, where the crosswalk was needed, we would focus on the vendor distribution.
That may be a long explanation for why we did this, but basically in order to build the website, we needed to make sure the two classification systems spoke to one another. The vendors used their own classification systems. As I think you will hear tomorrow, Doctor First, AllScripts, each vendor picks his own classification systems. So when you send the data out to the vendors, they do not need the classification system we built. But it was necessary for the website, because we needed to present the data in a standardized format.
That is a very long and short of it.
MS. THOMASHAUER: Does that answer your question?
DR. COHN: No, actually, standards discussions typically are very long and confusing.
MS. LOWE: I have a bad allergy, so I can't hear anything very well.
DR. COHN: No, I think you represented it well. Let me just ask as a followup question about this formulary database, we have had a lot of discussions around this issue of benefits and formulary. Obviously they are separate, but sometimes it is hard to pull them apart. We have heard a number of different approaches having to do with getting the formulary down to the provider at the time of the prescription. Others note that just a formulary alone without some understanding of the benefits that are a part of that formulary are really not sufficient. So others are talking about sending it down through the X-12, 270-271.
Does any of this stuff make any difference in terms of your view of how all of this should play out? It looks like you have been working with this for awhile.
MS. THOMASHAUER: That is one of the reasons I mentioned that ultimately we agreed. You already can get it to the individual level, because you are working with one or two plans.
The simplest thing is to make it available at the plan level, but that doesn't include the tiers. So there is some level of information about what it is going to mean to the patient once they get to the pharmacy, in terms of their payment responsibilities. That is why it is important, because when the patient gets to that pharmacy, they don't want to be hit with a $300 bill and a big surprise.
So to the extent you can give them more information rather than less, it is going to be increasingly meaningful to them, and also to the physician's office. I think as we stated, the amount of time that goes back and forth between the pharmacist because of what the patient hears at the pharmacy, and the pharmacist bears the brunt of it, then it reflects back to the doctor's office, where the staff has to go back and forth.
Our experience tells us the same thing; the physician has no idea how much time his staff is spending on this a week. So the value isn't recognized or appreciated, I think.
The greater amount of detail you can get to that patient, the better value for the patient and for the process. So if you go in and sign your name and they say to you -- they can look at your exclusions and they can look at your tiers and they can say, that will be $35, rather than saying it is on the third tier, that is more meaningful to you. So the better the granularity of the information, the more meaningful this is going to be.
Having said that, even the plan level is helpful at this point. The reason that Robert has had better success getting to the eligibility and individual level at this point is because he can decide within his own plan to work with his IT department to make that information available. When you are working across plans and you have got to confront every health plan and all of their priorities, as you know, the more IT changes, it becomes a lot more complex. So to get all the plans in the country on board at that level is going to be a huge investment of time, resource and commitment. That is why it is going to take longer.
MS. LOWE: I wanted to add to that, too. The vendor systems as I think you spoke to in Rhode Island, you only have one vendor system right now. Then each practice uses its own practice management system. So once we do have the patient specific information, it will be necessary to not only get the PBMs and the health plans to build to that patient specific level, which is quite costly, but then also have the vendors and the provider practices build their systems to accept the information.
So you are asking all the different levels that are coming through to invest significant resources to build to a standard that they may or may not think lasts. People have been unable to build to certain standards because they don't think they are going to stay around for too long. As we have all learned, someone said earlier, there were 129 companies, the person from the Rand study, there were 129 companies they started out with. Probably when CAQH started, there were 129 companies, all with different standards. So people have been very hesitant to build anything, since it is so costly.
Once a system is built, it will significantly help the adoption. I think you are finding that in HIPAA also, probably.
DR. COHN: Yes. I think a lot of what this committee's work has been, and subcommittee, has been to try to lessen the risk of implementation as well as development, by trying to show low risk avenues and approaches. So I think that is what you are reflecting.
DR. MANDEL: Let me make one correction. We actually have worked through RxHub. That is how the member specific information gets displayed to our physicians. That is able to display both, because we both use PDMs that participate with RxHub. So that is how our information gets out, but there are other ways of doing it.
DR. COHN: One final question. I know Jeff has a question.
MS. THOMASHAUER: Can I just say one other thing on this? The other thing that we are learning, and I think it is probably evident, but I just want to state it, is that to the extent you have standards, and it doesn't cost all the different vendors more money to interact with all of the different sources for the information, we are going to hopefully bring down the cost by technology to the user and to those physicians who can't afford it today. So I think the standards will also go a long way to encourage vendors to be in that market.
MS. LOWE: Right, it goes back to the chicken and egg scenario.
DR. COHN: Just to make sure I understand from our formulary discussion, I think what I am hearing is, you hate to use the term tiers, because that has a different meaning in the conversations we have, but there are different levels of this formulary benefit discussion.
I think what I am hearing from you is that at the very least, it would be nice to be able to provide the capability and the support to enable at least at the plan level formulary information. What you would really like to do, however, if you can't even get there now, at least move to this issue of being able to deal with it at the individual beneficiary level.
DR. MANDEL: In our incentive program that I talked about, it only applies to physicians who use technology that provides member specific formulary information. And basically, we have a number of different benefit plans. We have a benefit plan that has a two-tier pharmacy benefit, and we have plans that have a three-tier, and with different copays.
So the physician to understand whether the patient has a two-tier benefit or a three-tier benefit, he needs to have the member specific -- or, it is really contract specific, but basically member specific information, so he understands what the patient's benefit is.
DR. COHN: Let me ask another question. When you were talking about plan, I thought that was the level you were talking about. I guess I thought there were three levels. One is general plan, the other is your specific contract, and then the third is, you have already used up your deductible, you are over your max and therefore the cost is this much.
MS. THOMASHAUER: There are really two levels you need to worry about. One is the plan level, and then you get to the group or contract level, because for the most part you don't get the individual if you have got the group level. So really, what we talk about as the desirable state is the group level. At least, that is what I am talking about.
DR. MANDEL: And even though we have those different benefit plan designs, two tier, three tier, the same formulary applies. So any drug that is on the formulary will be on the formulary for all of the plans. It may costa different amount, based on how it is structured within that plan, but you aren't going to get the issues of, this isn't on the formulary, you can't have this medication covered by your benefit. You will at least know what drugs are covered and what drugs aren't covered. You may not understand the impact to that specific member, in other words, does he have a five dollar copay, $20 copay, $50 copay for his plan, versus somebody else who may just have a five dollar or a ten dollar.
Basically, the three-tier system allows us to provide more member contribution to costly drugs, typically multi-source brand drugs. So for members who have a three-tier, they are obviously going to have an incentive where they would like to have their physician prescribe the generic, which will be the least costly for them, or a preferred brand, and then there is the most expensive drug. So they would like to stay down. For two tiers, there are only two copays, so it is divided differently.
But the drugs that are available to the member that are covered by us as the insurer are the same. They are just maybe categorized differently in terms of the amount the member has to contribute for that drug. So if you get plan information, the physician will at least know, I am not going to get a call from the pharmacy that says the patient can't have this drug, it is not covered by Blue Cross. He may get a call that says, you chose this drug, it costs $35, if you use this one it would only cost five dollars.
So he may not get that with just plan information, which he would get with member specific, but plan is better than nothing, is what I am trying to say.
MS. LOWE: It is the first step, and many vendors at this point aren't even using the first step, so to start building to the next level is definitely a long term goal.
DR. COHN: It is an issue of standard support to enable the functionality. Jeff, I'm sorry.
MR. BLAIR: No problem. Robin, you got a lot of attention here. A couple of questions. You indicated that because of the lack of standards, you had to create crosswalks, so that you could reconcile or do comparisons between the drug knowledge bases. This was probably at a time before the National Library of Medicine, at least we think at this stage, is looking at accepting the CHI and NCVHS recommendations to use Rx Norm as --
DR. HUFF: That is not a classification.
MR. BLAIR: Pardon?
DR. HUFF: That is not a classification. NDFRT is a classification.
MR. BLAIR: So it would be the representation of mechanisms and actions and physiologic effects, mutual suitability and all that stuff. So in short, I am wondering whether -- are you familiar with those plans?
MS. THOMASHAUER: I am not.
MR. BLAIR: Okay, so let me put that question completely aside. The other pieces are, you had to develop some ways to communicate information on the formularies. Are those like message format standards that you developed, or what?
MS. THOMASHAUER: We developed data element standards. There were certain pre-agreed formats the plans could send them to. So they had an option of how to do that, but the data elements were what we standardized.
MR. BLAIR: Were they similar to what RxHub had also developed?
MS. THOMASHAUER: Yes.
MR. BLAIR: RxHub indicated earlier that they were planning on taking a number of their proprietary communication standards to contribute into the Open Consensus Group's standardization, like NCPDP script. So the investment that you made will become available to everyone as a consensus based standard.
DR. COHN: Are you referring to the mapping at this point?
MR. BLAIR: They are two different things. One is the mapping and then the other one was the communications of the formulas. The second one was the formularies.
MR. REYNOLDS: To David, Robert and Laura. As you talk about the implementation in the physician's office and the business process re-engineering that has to go on, what percent of your scripts are you seeing that are being done through what you would consider a fixed desktop, in other words, by some admin person, a laptop versus a PDA? Is it because of the amount of data they have to deal with?
When we talk about that technology, it is easy to say PDA or easy to say this or that. What do you actually see on the ground when a physician is talking this along, how does it work?
DR. ADAMS: For us, the answer is relatively easy in Rhode Island, because with our vendor, their particular product sits better not on the hand-held, so we have almost none being done through hand-held because of the vendor we are using. They are working to have it work better on hand-held. Doctor First works tremendously well on hand-held because it was designed for that, but we only have one or two people using Doctor First in Rhode Island practices, so for us it is de facto the desktop.
MR. REYNOLDS: Does the doctor do it or does his staff? Does he still write the script or does the staff do it?
DR. ADAMS: That wasn't going to be allowed in Rhode Island, so we don't have that capability with it. That is not allowed. The physician either will write the script out and hand it to the patient, or they put them into the system. There is no staff member taking the script from the physician and putting it into the system.
DR. MANDEL: Typically we find that the physicians are using the hand-held, and their staff are frequently using the browser version. The staff will do that. We are not hooked up to SureScripts yet, so the renewals are being called in, and frequently the staff will queue those up for the physician, so that the physician can then go in on his hand-held and check, yes, yes, yes for renewals. So that is typically where the -- and the browser version is typically used to enter the demographics for the new patients.
DR. MEDVEDEFF: Everything we are doing is on a PDA, and the physician is the only one entering prescriptions. Our authentication process, we are using the AMA Internet ID through Verisign, which is a PKI technology. So the only people logging into the PDA is the physician.
MS. LOWE: For the DC Pilot, we use both desktops and hand-helds, just because we have been told that you cannot select the type of mechanism the physician is going to use to transfer the data. So we did not try to make a decision for them.
MR. REYNOLDS: Would you recommend that the standard be general enough that it didn't really focus on where it ended up?
DR. MEDVEDEFF: Yes.
DR. ADAMS: Yes.
DR. COHN: Other questions or comments from the subcommittee? I want to thank our presenters. It has been a wonderful set, and I'm sure we will be following up with you with additional questions as we move along.
I suggest we take a 15-minute break, and we will come back at 12:30.
(Brief recess.)
DR. COHN: The next period is an open microphone period. It is an occasion to invite anyone from the audience who wishes to make a statement or a comment, either on the events of today or generally on e-prescribing, to come forward and comment. If you like, you can come to the front microphone. Give us your name.
PARTICIPANT: I just had a few observations I wanted to share, around two categories of things. One is thinking about scope. A lot of interesting discussion today, but I am a little concerned about whether a process for creating standards for electronic prescribing necessarily can solve all the problems of the health care system. So we have to be careful not to take it too far into EMR or too far into requiring things that the infrastructure doesn't exist today to support.
For example, there was a lot of conversation about getting to actual cost for a member. Today what we support is group level information, but when you start talking about actual cost, it goes eons beyond that. In addition to getting information about the plan and the benefit for this person, you have got to know what pharmacy they are going to. You have got to know that that pharmacy participates in the network, and you have got to know pricing from that pharmacy. So getting to actual costs for a specific drug is a significantly more complex question than getting to relative cost, which is what we have always talked about.
If you talk about three-tier benefits and so forth, you can say the third tier might be $30 and the second tier might be $20, and that gives people enough indicator, or if it is non-formulary, it gives them an indication that they are basically going to be paying for it, even if it is at some discounted rate. But to say you want to try and get to actual cost is much more complicated. I just wanted to make sure people were aware of that.
There was some conversation this morning about drug history. I think Jim Bradley made a comment about what we call shoebox claims, which are claims at the pharmacy where the copay is actually higher than the actual cost of the drug.
I wanted to clarify. It really isn't a big problem. Our pharmacy contracts require the pharmacies to submit a claim, but even aside from that, for the pharmacy to know what the member responsibility is, the way the NCPDP system works is, they submit a claim. So for them to even know that the cost of the patient is seven dollars and the copay is ten or something, the claim comes to our system. So I don't want people to get the wrong impression about the level of drug history information that exists, because it really is pretty good.
The problem with OTCs is different. I think you have got to be careful not to try and solve a problem where there is no infrastructure to support it, because if I buy Claritin or if I buy Prilosec at Sam's, there is no infrastructure for them to collect my name and address and put it into some kind of a system. The same thing is true at Walgreens or CVS. You take that to the front counter that is not at the pharmacy counter.
So I don't think this process is going to allow us to get to solving the entire drug history problem. I think that is worth keeping in mind.
A lot of these come down to creating a standard that allows for perfect, but provides for good enough. I think right now in a lot of these cases, with cost information and with drug history information, there are a lot of systems that exist today that get us pretty close and provide a lot of value.
I made a comment the last time, I think there is a risk here that if you try to solve too many problems, you get to a point either that the transactions will take too long because they are bouncing off too many things, or they are trying to create systems that don't exist, or will create costly systems that physicians won't pay for, or will just create so much resistance from different places in the chain, because you are forcing people to do things that they don't have the financial resources or the incentives to do, that it could kill the whole thing.
So again, I just want to make sure that we keep the scope within reason, and basically provide a standard that takes into consideration what exists today, and doesn't try to go beyond that.
DR. COHN: Questions or comments from the subcommittee? Next testifier?
DR. BURNS: I'm Terry Burns from RxHub. I want to give you a little bit of my background so you know where I am coming from. I have two things I want to comment on.
First of all, I have been at RxHub for two and a half years. My job there has been to look at existing standards, determine where we can utilize them to do what we needed to do, as well as develop standards where they didn't exist. I ran all the work groups and got all the right people in the room, helped develop the standards, got comments on them, implemented them, piloted the process, et cetera.
That is what I wanted to talk about. I and a lot of other people -- it was a very collaborative effort between PDMs, pharmacies, payors and technology vendors. I just facilitated it. But just so you understand the background and where I am coming from, I think it is important to note that whether our standards -- there may be HL-7 standards and NCPDP standards, and we need to make sure that we adopt them both if they exist, because we can't force hospital systems to use NCPDP, and we can't force pharmacy systems to try to use HL-7.
It is important that we have a collaborative effort to understand where the data can be transferred between both, and RxHub has done that. We use both the HL-7 format for medication history, and we developed an NCPDP-like one for medication history to utilize with the PDMs, because that is what they are used to, and the technology vendors.
So you may need to have networks in between that can translate the information, as long as you don't force the tech vendors to have to redevelop everything that they have done and adopt a new standard or force the hospitals to do the same thing. In addition, you may have a tech vendor that uses it both in the ambulatory setting and in a hospital, where they are communicating both with a retail pharmacy and an in-house hospital pharmacy, and they are going to have different standards that they utilize between those two.
So that is my spiel on standards. Then I wanted to give you the rest of my background. I worked for a PDM for seven years, so I wanted to help you understand the benefits, the formulary level. I think what is important to understand about that is that different plans have information at different levels. So one plan may have one formulary, another plan may have five formularies that they want to communicate. One plan may have 12 benefits, another plan may have 250 benefits.
So what is important is that you can transfer that information to the technology vendor, that they have all of the information that they need, and then somehow you can tell them how to tie the patient to whatever information you have given them. So it doesn't matter if it is at a plan or group level, if it is contained and it is identified by some key. Then when you do the eligibility transaction, you tell them what keys to use to access that information.
So whether that is downloading information to the technology vendor's application or whether it is a real-time transaction, you give them the keys on an eligibility transaction and tell them where to find the information that they are looking for. Whether at a plan level or a group level, it doesn't matter. From a standards perspective, it is finding the information and having a standard way to communicate the information at whatever level.
DR. COHN: I think there is a question. By the way, thank you for the discussion about the formularies.
MR. BLAIR: Thanks for those comments. Has RxHub created some mappings between some of the HL-7 messages or data contents and those within NCPDP script? And if so, is that something that should be preferred and considered within the consensus based standards community?
DR. BURNS: We haven't done it for new prescriptions or refills. What we did do was, originally we created a new transaction utilizing the script segments for medication history. it is our plan to take that to NCPDP and have it approved as a new script standard. But then we found that when we started working with the hospitals, they didn't talk in NCPDP.
What we can do is, we can take an HL-7 request from the hospital vendor, we can send an NCPDP request to the PDM for the information, and we can translate it back into an HL-7 response to the hospital vendor. We have done that for medication history. There are industry efforts going on to do that for script as well, between HL-7 and HCPDP.
PARTICIPANT: And as you heard June testify earlier, whatever standards get developed, RxHub is planning to support them all the translate between them. So I think that is important to be able to do that.
DR. COHN: Thank you very much. Other testifiers?
DR. MARTIN: Good afternoon. My name is Ross Martin. I'm from Pfizer. I'm a medical informatician. I participated in many standards organizations that participated both last time and this time, and know a lot of vendors that have talked today.
The questions that I have that were stimulated by today's conversation, which was quite fascinating, revolve around how the payors -- let me back up. In NCPDP, as a representative of a pharmaceutical manufacturer, we have an observatory position there, because we don't actually do direct transactions. One of the things that I observed in that environment is a lot of discussion about the burden of who is going to have to collect information, who is without -- a payment without representative kind of thing, or taxation without representation.
I see both sides doing this between the payors and the pharmacies, saying the payors are asking us to provide a patient weight at the time of prescribing, and what are we going to do, put a scale next to the pharmacy kiosk and have them step on, Mrs. Jones, and let me provide that for you. These kinds of conversations go on all the time.
I would suggest that you couch some of the comments that you are getting about how hard certain aspects of this is from different players about, it is hard for them, but what is the ultimate benefit, specifically around providing payor and eligibility information. The question becomes, what is the ultimate benefit to the payor or PBM of providing certain levels of information.
It is not surprising to me to find that the level that seems easiest to provide is the one that is most directly beneficial to the payor or PBM, as opposed to what the patient really needs at that point of care. We all know from the Medicare legislation how complex the payment structure is, or the suggested payment structure, which can be changed by the individual PDPs, or prescription drug plans.
That is the information that the person, the patient, really needs to be able to determine what they are going to do in that clinical situation. I need to make a decision about what I am going to pay for this drug, and I need to know how far into the doughnut hole I'm going to get if I am going to be on this chronic medication. Fine, if it is covered at this time with this particular copay, but what is going to happen when I am on this for three months, and I find that now I am in the doughnut hole and I owe 80 percent of it, or whatever that plan is going to be.
So I completely support the idea that this is a tiered process, in the definition of tier that we have been talking about, not tiered like tiered formulary, but tiered implementation strategy. This is hard to do, and as you go further into it, it gets harder and harder. But you need to think about, in each of those phases, what is the most important for that patient, and what is going to provide the maximum benefit for them at that level of complexity, and stick with that, and not look at relative hardship for any individual player within this.
We obviously have to work it out, but I'm just concerned that that conversation be part of the decision making process of this body.
The other comment to make is about the last series that we heard, of the levels of passion of these players. I think that is a critical element to all of this, that as we do combine and seek to collaborate, as Laura Adams was suggesting, we are going to continually require people who are linchpins to make all this stuff glue together.
The last comment I'd like to make is about the therapeutic classification scheme. I heard you asking questions about that issue, and I am not sure that they were adequately addressed by any of the speakers. I think it is a challenging issue. There is still not clarity about how we are going to represent therapeutic classification by vendors, by PDPs, and are they all going to be different, is there going to be opportunity for there to be a lot of differences in those classification schemas, is there going to be one classification schema that different PDPs select different levels of coverage, in terms of lumping things together or splitting them apart. Those things need to be figured out.
The reason that I think it is important for this body to consider is because there are technical implications to that. I don't know how the vendors are going to address this issue without creating some sort of crosswalk that is universal. The discussion that Robin was having about their efforts to do it between two different -- First Data Bank and Medispan, two compendia, how complex that was, and they ultimately decided this is really hard. What is going to happen if we have 26 different therapeutic classification schemas, which is a very possible reality within the current legislation, as I have been able to read it, and anybody I have talked to about this as they have been able to define it, or not. So I'd like the NCVHS subcommittee to think about that issue as well.
PARTICIPANT: I'm just a little puzzled what you mean by the therapeutic classification scheme. I can understand why CAQH did what they needed to do, because they were trying to bring a lot of different players together. But at the end of the day, if you are able to provide to the physician the information they need, that clearly for that particular patient is very specific to what the drug classes are, and provides the right information that the payor, whoever they may be, has designed, I guess I am just not understanding where you need to go and what is missing with the therapeutic classification scheme, even if it differs by different benefit manager or health plan today, why that would be an issue at the end point between the doctor and the patient as they make their decisions.
DR. MARTIN: I think there are a couple of reasons why it is confusing. I think it is going to be confusing on a patient level to even know what the benefit is, if there are a lot of different classification schemas out there. It will be hard to know how it applies to them individually, one, if they can be different and two, if they can change with some frequency, which is the way the legislation is currently written.
I think it is confusing for the provider, the prescriber, to think of things in a certain way, clinically, about classification, and then see it presented in a different way within the prescribing environment from payor to payor, and not understand that it is a mechanism for managing benefit. I think that is going to be confusing for them.
For the vendors, the vendors that I have talked to are going, we're not sure how we are going to address this. This is an issue for us.
PARTICIPANT: I guess it is a potential issue, and it is maybe something that a little bit more research and work needs to be done on it. From what I have seen in the e-prescribing arena today, I wouldn't say that there has been real problems with the way things have been classified. Usually particular products fall into the same therapeutic categories over and over again, and usually there hasn't been too much of a major issue, but perhaps that is an area for some work. Maybe we will hear some more from the point of care vendors who are going to participate tomorrow.
DR. MARTIN: I think that would be very helpful. That is one of the questions I want to see addressed. I think the reason there hasn't been a lot of confusion is because predominantly, the vendors tend to glom onto one or the other of the major compendia, so they are able to use that as their framework, and there hasn't been that much complexity so far.
This bill is potentially introducing a lot of complexity that doesn't current exist, and that is part of the point.
PARTICIPANT: I would agree with you. I'm just not sure that complexity is necessarily related to the classification schemes as it is to everything else.
MR. BLAIR: One thing I need some help on, and I thought that I understood what was being referred to with classification schemes, but as the conversation goes on, I'm not sure I do. What are we talking about when we are talking about classification schemes?
DR. MARTIN: Within the context of MMA as I understand it, and I wish I were more knowledgeable to the level that I would love to be able to say with some authority, but I think it is because there is a lot of complexity about this.
I hesitate to answer that question directly, because I see five people in the room that know the answer better than I do, and I am wondering if they want to field it. Mr. Bazaro from First Data Bank will be able to address this very effectively.
MR. BIZZARRO: I am Tom Bizzarro with First Data Bank. Also representatives back there from MediSpan, and George Robinson from our editorial department is back there also, so if I mis-speak, they can help me.
The therapeutic classification systems that we have in the compendium right now are unique and proprietary to those compendia. They are meant to group drugs together of similar therapeutic classes. The specificity within those therapeutic classes is different between the different compendia and between the different classification systems.
I think that Ross' point is well made. With my understanding of this legislation, drugs within certain therapeutic classes need to be covered by the PDPs. One of the indicators for coverage of the drug is, is it in a particular therapeutic class level, then that can make this a very complex situation if there are variable classification systems out there.
The other thing that concerns me a little bit is, do the manufacturers and the PDPs now come under pressure to try to get drugs in a certain therapeutic class, because they know it is going to affect the way that drug is reimbursed.
One other thing I want to clarify is that it is a therapeutic classification system, it is not an indications database. The therapeutic classification systems should not be used in that manner. They are therapeutic classification systems for the products, not an indication database.
DR. COHN: Let me ask a question then, since we have you here. We will probably continue this tomorrow, I would imagine, but I had thought there were two things happening that may in some way mitigate the concerns that we are hearing.
One is that I had understood that the USP was in the process of coming to some grips with what will be the classification at least as used by Part D. I presumed, and maybe I am mistaken, that that would include an identification of the drugs in those classifications. Part of the rules is that you have got to have a certain number of drugs from each classification offered on your formularies.
So I guess I thought that that was a piece of the solution. I also thought that another piece of the solution was the work I thought was going on from the National Library of Medicine relating to NDFRT and mapping to what I thought was basically currently existing proprietary classifications. Am I off? Did I misunderstand?
DR. MARTIN: I think it is a partial understanding. This is one of those subjects that is open to a lot of interpretations. What did the writers of what is now the act, the MMA, intend?
My understanding of the history of the process was, there were two competing legislations, the House bill and the Senate bill, and this is how they ended up amalgamated together. One bill said the government will provide a classification schema. Presumably it would have been something akin to NDFRT. The other bills said the payors would each create their own classification schema, and this turned out to be the compromise.
What the compromise says is that USP was requested by Congress to create a standard classification schema. I don't know if it was quote standard, but it was a classification schema that may be adopted by the PDPs. It was definitely may, not must. They don't have to take that classification schema and use it for their classification schema. It is the suggested retail price, if you will. It is the suggested classification schema. If they don't adopt it, they can create anything technically that they want.
Now, I don't think it is in anybody's best interest for them to do that, but the bill does not say you have to agree upon one schema. I think it could create a lot of confusion if it is not done in a consistent way. The motivation, as Tom was saying, is, if you are a payor and you have to cover drugs in a class, you are incented to lump things together versus not.
DR. WARREN: I have a question for you. My first is trying to understand your concern on the complexity. I think one of the concerns is, if we have competing schemas, then we have problems with decision support being provided at point of interacting with all of this stuff.
DR. MARTIN: More about understanding the benefit. I think the compendia, like the First Data Banks and the Medispans of the world, will be able to provide decision support based on their own compendia that is not tied to the benefit.
DR. WARREN: It is when we start sharing that, it is an interoperability issue.
MR. BIZZARRO: We need to be careful when we talk about what we can do in the compendia to provide decision support and therapeutic classification systems, because those therapeutic classification systems really don't have a direct impact on how we do our decision support.
I guess my concern here with the classification system is that, there are seven or eight of them out there right now that I am familiar with. All of them have different levels of specificity. Each of them have drugs at these different levels of specificity that you then can roll up in a hierarchical manner. So if you come up with a therapeutic classification system that is hierarchical, at what level do you assign drugs that are going to be included in a benefit.
So not only is it just a therapeutic classification system, it is how drugs are assigned in those different levels of the class.
DR. WARREN: My second question was a really dumb one, but I think you just answered it. If I am used to a therapeutic classification, this is an antibiotic, this is -- okay, and what you are saying is, you are trying to classify these according to how the plan is going to provide these drugs, for what purpose.
MR. BIZZARRO: I'll give you an example with antibiotics and just give you a possible hierarchy for a therapeutic classification system. Anti-infectives, antibiotics, beta lactin antibiotics, sulfalexin antibiotics, first generation cephalosporin, second generation, third generation, fourth generation. So the drugs of the fourth generation fit into all those other levels except for the first, second and third, so where do you classify that drug. A fourth generation cephalosporin is going to be an awful lot more expensive than a first generation cephalosporin.
DR. WARREN: So you are classifying the way the drug interacts, but also for the cost? I am trying to understand why you have a different classification system than what I would see if I went into a pharmacy textbook or something like that.
MR. BIZZARRO: The cost has nothing to do with the drug as far as we assign it to a classification system, therapeutic classification system. It really has to do with the pharmacology of the product, and how our clinical editors see that drug fitting within the hierarchical classification system. What has been demanded of the industry, of the compendia, more recently is therapeutic classification systems with more specificity, because as drugs get more complex and they have more ways to define them, you want to put them at a level that is appropriate for that particular drug.
You also need to be able to roll them up in this hierarchy, but you also want the specificity. It is a very complex subject, and I hope I'm not getting too far off track.
DR. WARREN: I think I understand now some of the complexity.
DR. COHN: From a physician's viewpoint, if you are a single payor, it is obviously not much of an issue. But if you have got four or five different PDPs, all of which are calling their things different things, and you are navigating this large drug list and you are used to seeing things, anti-infectives, or maybe in one classification all the cephalosporin are together and others they are separated, I don't think you are going to be able t find anything on your list, because they are going to keep changing by the formulary. isn't that the outcome of this? It is called massive confusion.
MR. BIZZARRO: It is absolutely true. Even though we develop and continue to develop more specific classification systems, many people want to use the old less specific classification systems because they don't need that level of specificity for whatever their business need is.
The other thing I want to mention and make sure it is on the table is that if another national organization comes up with a classification system that is decided as the one we are going to use in the program, someone still has to take each of those individual drugs and make sure that they are placed in the proper spot within that therapeutic classification system.
So USP developing that classification system, that is one thing, but how do we assure that the different compendia or whoever is doing this, whether it is the compendia or the PDPs, put the drugs in the right place in that classification system.
DR. COHN: Maybe we can hold this bookmark also to have the NLM come back and explain to us all this.
MS. HOLLINGER: My name is Kathy Hollinger, and I am from the FDA. I have been working on a research project looking at the terminology that is used in package inserts. We were looking at some of the language that is used in the package inserts to characterize drugs when they characterize groups of drugs, terminology like psychogenic agents or neuroactive agents. So we found a lot of terminology that groups drugs, based on a number of different factors.
When you gave a list of your classification system for antibiotic, what I heard was a therapeutic intent for antibiotic, meaning a therapeutic intent. But then when you went lower down in your hierarchy, you actually were talking about chemical structures.
So what we found in looking at that terminology is, there is not just one classification system, there are probably several, and they are probably based on things like mechanism of action, physiologic effect, therapeutic intent, because there are many different ways you can classify agents.
Being a veterinarian, we also use drug off label a lot, so if you use therapeutic intent, it would be improved for certain therapeutic intents, so you couldn't use a therapeutic intent classification system to look at drugs that you are going to use off label. You might be using physiologic effect.
So I guess I just want to convey that I think you will end up with more than one classification system potentially.
MR. BIZZARRO: I think you made a good point for why I didn't want people to look at this therapeutic classification system as an indications database. In an indications database you can have both indicated use, on label use and off label use for drugs, and you can document what the level of documentation for the use of that product is. So they are different things, absolutely.
The other thing which you mentioned is that we are market driven development of the classification systems for the level of specificity. So if we have customers who need that distinction of a drug product at a lower level than you may think is necessary in a true therapeutic classification system, we will take it to a lower level of specificity. That is what we heard in the marketplace.
MS. HOLLINGER: Maybe then again the payors will have a different classification system themselves as well, and be able to map to different ones.
MR. BIZZARRO: To tell you honestly, and I don't want to speak for the payors, but the payors have been dependent on the compendia to do that because of the complexity, not of developing the classification system, but the ongoing maintenance and the addition of drug as they come on the marketplace as drugs change, and then changing them within the classification system.
PARTICIPANT: I agree with Donna. I think this is an interesting issue. It is an issue that CMS needs to work out as far as mandating to the PDPs what they are going to cover under the MMA. I think it is beyond the scope of creating standards for electronic prescribing, because I don't think with respect to electronic prescribing this is an issue or will be an issue.
I think it is an issue for Medicare and what Medicare is going to pay for, but I think that is beyond the scope of this group.
MS. BURNS: Terry Burns again with RxHub. I just wanted to clarify something for you, Simon, because I don't want you to leave with the impression that the tech vendors are displaying all these different classifications to the doctors; they are not.
What happens is, we create a standard of information mostly at the NDC level, so that the tech vendors can then roll it up and classify it however they want, depending on the purpose they are using the application for. So the tech vendors are choosing their source for --
MR. BLAIR: What are tech vendors?
MS. BURNS: Like a Doctor First, or anybody who is writing the prescribing application, or developing a prescribing application. They are choosing a source of data, and they use that source of data pretty consistently. So they aren't seeing all these different classifications, the proprietary ones in First Data Bank and Medispan.
I'm not saying this is not an issue, because I believe it is, but the issue is how do you transfer the data so that they can roll it up and use whatever scheme that they want to use. It is probably something you should ask them about tomorrow.
DR. COHN: I hear what you are saying, and I guess we were just following the logic of getting the formulary for the members' plan down to the provider, so that the provider knows what is covered, what is not, what is available, what is going to be very expensive and what is not going to be very expensive.
So yes, it is great to have a vendor give a standard list of drugs, but I don't know if that is quite the same as a formulary, unless you are only dealing with one pair.
MS. BURNS: The way we understand it, having developed that standard, the formulary standard for them, we actually allowed the PDMS to send the information at the NCD 11 level, so that they could get down to the lowest level that they needed to in order to communicate the information. Then the vendors use that information in whatever ways appropriate during the prescribing process.
So if they understand that a particular type of package or a drug isn't going to be covered but another package would be, they have that information. However, they don't want to display all of those NDCs to the doctor, so they may roll it up for the doctor as he is trying to peruse the application during the e-prescribing process.
So I don't know what levels are used, and like I said, that is a really good question to ask the vendors tomorrow.
DR. COHN: Thank you.
DR. MARTIN: Apologies to the committee for introducing that bailiwick there.
DR. COHN: We will keep it in mind tomorrow as we ask others what is going on. Any final comments from any of our consultants?
DR. ROBINSON: I'm George Robinson with First Data Bank. I think Tom said most of what I wanted to say.
The issue that I see, with the USP jumping in and assisting with the formulary, is that the USP will be establishing some therapeutic classes. We have National Library of Medicine creating the clinical drugs for Rx Norm. We have the FDA providing ingredients for the unicodes. We have the NDFRT providing physiologic effects, and the layers that we are starting to get are very complex as far as what entity is going to maintain the relationships on an update cycle that is usable at the point of care.
I'd like to know more about the USP initiative, because as Tom was saying, one of the greater challenges is interpretation, especially if you look at a multi-ingredient product; does it fall into a general cough and cold category, or does it fall into an allergy category, and on that type of delineation itself may include or exclude payment of a drug in one formulary or another.
Those are the types of calls we handle every day at First Data Bank. Tomorrow we will get into what we think is the role of the drug knowledge in order to help with some of this, but it is still where the associations are going to be made between all of these relational components of defining the drug. I don't have a good handle on that right now.
DR. COHN: Thank you. I guess it is always helpful to look under rocks and discover what is there.
DR. ECKERD: Karen Eckerd with MediSpan. One of the questions you had asked is what we have provided to UMLS for mapping purposes. From the MediSpan perspective, all we have provided has been our proprietary version of the clinical drug. We have not provided any of our therapeutic classification systems for mapping.
George, I don't know if you have done differently or not, but in lieu of UMLS not being here to help answer that question.
MR. BLAIR: I think that was understood when the NCVHS wound up separating the portions of NDFRT so that they wouldn't compete with the private sector drug knowledge base sectors. That was why we only specified the representation of mechanism of action of drugs.
DR. ROBINSON: Yes, that is correct. My understanding, where NDF stops in the FRT is at the clinical drug. Then I understand there will be a relationship between the clinical drug from a generic to a brand, and then between those clinical drugs to the NDC. I understand that is in the purview of the National Library of Medicine.
Our therapeutic classes are made available to help the National Library of Medicine determine where a drug should be handled. But it is not -- the therapeutic class itself is not being published within the Rx Norm.
DR. ECKERT: The physiologic effects and mechanisms of action that are being made available in the NDF RT are a portion of a therapeutic classification system. So we can provide additional information to map to, but it is still not going to completely answer entirely the therapeutic classification mapping situation, because there are other things incorporated in therapeutic classification structures than just those two pieces. Thank you.
DR. COHN: I feel that we have gone all the way around on this discussion. My own view is that there is probably a question -- unfortunately Robin Thomashauer isn't here to ask her; if indeed all of this is the case, then why were they not being First Data Bank and Medispan, and have her again explain to us what they felt was the business case that made them invest what sounds like a million dollars into that mapping. So once again, somebody must have thought that this was an important activity.
I guess I am reminded that this is only -- maybe not quite handled by the NLM activity.
MR. BLAIR: One last comment. Are all of our testifiers here from this morning? They are not, okay.
DR. WHITTEMORE: I may be the only one, Ken Whittemore from SureScripts.
MR. BLAIR: Okay, that's okay. I was going to be addressing my comment there, so that is okay.
DR. WHITTEMORE: A point of information. I think I understood Dr. Mandel to say that -- quote a statistic that 50 percent of prescriptions weren't filled. We have looked at that, and the figure that we have come up with from a number of sources is more on the order of 20 percent. So I wouldn't want the committee to walk away with the one impression there. That is all I had.
DR. COHN: That is the number we see from many sources. Other comments, questions? That was a very interesting open microphone. I like this.
MS. CANFIELD: I am Anne Canfield with Rx Benefits Coalition. I just wanted to clarify one thing. There was some discussion this morning about pre-emption or state activities. The language in the legislation is extremely broad. It was a specific pre-emption provision put into this particular section of the bill so that what is happening at the estate level is irrelevant because the pre-emption is so broad.
DR. COHN: Any other comments? Any comments from our subcommittee members?
MR. BLAIR: The first thing I want to start off with is to thank Maria Friedman again for pulling together this agenda. I think that the testifiers today were extremely helpful in helping me, and I assume other members of the subcommittee, get a much better understanding of the environment and the context of the standards, as well as what standards are being used, and the perspectives especially that the network providers had with respect to what is a standard and what is not.
One of the things that I found extremely helpful in preparing for today on my behalf is, I read through two of the three testifiers who are network providers, filled in the written testimony, the spreadsheet, as to what standards they are using, what terminologies, identifiers they are using, and their feelings about them from a strength and weaknesses standpoint. I would encourage the rest of the subcommittee members, if you haven't already done so, if you have a chance tonight before tomorrow's testimony, there is written testimony from the software developers that we are going to hear tomorrow morning, and that is also very helpful in helping you feel more comfortable with that testimony. So I would encourage you to take a look at that.
MS. FRIEDMAN: I would like to encourage everyone to fill in that grid. It is on the NCVHS website. We recognize that our time is limited, and unfortunately we couldn't have everybody testify. But we are very much interested in peoples' input and perspectives.
So if you weren't invited to testify but you would like to fill in the grid and send it in for our consideration or for inclusion, we would appreciate it. The instructions are there on the website. You can also e-mail it to me at mfriedman@cms.hhs.gov, and you can snail mail it to me as well. The address is there.
MR. REYNOLDS: I'd like to thank everybody for their presentations today. It is different from having sat in the ring for HIPAA. The industry seems really interested in making a difference. They seem really interested in building on whatever comes out of this. There is a real positive energy, until we got to therapeutic codes. As a layman, I guess I will catch up tomorrow. But I really think that is a different area than some of the other standards that we have been talking about.
DR. HUFF: What occurs to me, and part of this is motivated by what I am hearing outside, not necessarily what we heard today. That is, a couple of weeks ago I was in a meeting with Secretary Thompson, where they announced the new EMR czar, David Breyer. His comment then was, he would like to see these things happen a lot faster -- he was particularly talking about e-prescribing -- than their legislated time line.
It occurred to me in the discussions today that there are some things that are non-controversial, that we can decide now or soon, or maybe everything needs a little work, but there are some things that we can decide a little earlier, and there might be a lot of benefit to staging this, and for instance, just as a straw argument, we adopt early and say EP script is what we would use for the prescriptions, and say that early and let people go on and work with that part, while we continue with hearings related to other less understood or more controversial proposals about that. Maybe that is already the work plan, I don't know.
The second thing is just a way of organizing and thinking about that. In looking at this, basically the idea is, how do we organize what recommendations we want to make. This is an organization that stems directly or very close to what was in the worksheet that Jeff sent out.
Some of the things in the worksheet, they talk about them as functions, so it is different. They say there should be standards for drug interaction testing. I don't know what that means exactly. That is a function in software, so there are standards for the codes that would allow that. I guess we could make a statement that required that functionality to be in software, but there is no standard. I'm not sure what that means, to have a standard for drug interaction testing, drug interaction evaluation or something.
Anyway, in areas where we have standards, I think there are identifiers, messages, codes, rules and representations and classifications. That is a mixed bag of things. Functional requirements and digital signatures, things that I think we would clearly not do anything -- we are not going to choose a particular van or say that everybody in the nation needs to use this particular application to do their prescribing, or that every pharmacy has to use a particular system. What we are talking about is these other kinds of standards.
So within these other areas, I think what we are saying is that with identifiers and things I heard today, with patients, prescribers, these are all things identical to what is in the worksheet, identifiers for pharmacies, identifiers for health plans, PBMs, insurance companies, identifiers for locations, for facilities, clinics, that sort of stuff.
Messages, again, there are more details in the worksheets, but there are prescriptions, renewals, eligibility, formulary access, and there is prior authorization and medication history. But the idea of this is that if there is a way we could start with some kind of list like this and say what do we think is most important and what could we do early, I would be in favor of doing that.
DR. COHN: Did you want us to look at your list? Is this a suggested framework for us to think about?
DR. HUFF: Obviously I would be glad to share it with you. if we go back a level then, the general idea is that basically in the standards, we are considering identifiers, messages, codes.
DR. COHN: What happened was that you expanded and then you quickly started to scroll down. I don't know that anybody talked about whether there are other areas.
MR. BLAIR: Could I make a suggestion on this and try to put this in perspective? I think what you are doing is extremely productive.
What Stan is referring to when he referred to that work plan or something, what we did in trying to help the testifiers prepare to testify is, we did it in two stages. The MMA law had a long list of what they called requirements for standard, which when you start to lay them out, really were what I would call potential areas of standardization that they wanted us to explore or consider. Then it had a second list of things that I would call attributes for standards, such as patient safety and quality and whether it is the low cost and all of those other things.
So we prepared that, and that is the two Excel spreadsheets that were sent out to the testifiers, asking them to fill in and respond in written testimony, because they are too detailed for oral testimony. So we did a much shorter list for the oral testimony to get at the highlights.
I think what Stan is observing is a little bit what I expect, that a lot of the areas that the law wants us to look for are likely to turn out to either be applications or performed by applications, and we may not need a standard per se for it, or they may turn out to be things where things like express scripts created a standard internally, and maybe they will share it with NCPDP. They may or may not need to be recommended for national adoption. The fact that they are being used and these functions are being performed may be sufficient.
So I think what Stan is looking at now is, he is taking it the next step with the testimony we have got today and said, there is some low-hanging fruit here. The only thing, Stan, that I would suggest is that the testimony we have received today has been from the network providers and from the different locations that have been early adopters.
I think if you continue with the list and wind up coming to some conclusions after we hear from the software developers that are going to have a different perspective, and the private sector drug knowledge bases, which will have a different perspective, and then on Wednesday morning we are going to hear from the providers themselves via the professional associations. With those three new perspectives added to what we heard today, I think that the list that you are building probably could be a pretty solid first cut at the low-hanging fruit.
DR. HUFF: I agree with that. That is sort of what I have been doing. With this framework as I listened to the testifiers, I just said that is something I didn't have on my list. That is a kind of identifier that I didn't know about, that is a message type that is implied by that functionality. So yes, this list could certainly grow.
So I guess I am saying, do people find this as a way to organize thinking about this, and then secondly, it is really a second question, do we think there are things that would be useful to adopt early, or do we think we wait until August next year before we make any recommendations? Or what is the plan?
DR. COHN: Let me remind you about the plan, Stan. The plan is that we are actually going to be holding additional hearings in July and August, four more days of hearings, asking others to help us flesh out.
I think your framework is actually very good. I think right now, our plan was to come forward -- if we meet in September, we may be meeting in October, but the plan would be to come -- you agreed to this.
DR. HUFF: I agreed to that. That is what I'm thinking, that we could do something much earlier than that.
DR. COHN: We may be able to. We have already talked about interim recommendations for the November meeting.
MR. BLAIR: I think there will be final recommendations.
DR. COHN: The way we are going, there may be. But I think we need to see where we are.
MR. BLAIR: The other thing that I would add to this discussion if I can is, the activities that we have, the way they are laid out, is that in May -- which is these hearings today -- and in July and in August, we endeavor to try to hear all of the stakeholders that the law said we must listen to. We were going to gather the opinions of all of these users and stakeholders by the end of August and say, here is the summary of the issues and the gaps that we have found and what the law asked us to look for, and feed them back to the STOs and terminology developers so that they can testify to us in September and say, did they address these gaps, do they have it in their pipeline, can they address these issues.
Then, based on the feedback from the STOs and terminology developers in September, we were going to pull together our interim recommendations, which would be approved by the full committee in November. So we would have to wind up working on those recommendations during September and October.
Now, there is one major piece that is left out of that schedule. The piece that is left out of that schedule is electronic signatures, which we have scheduled for February and March, hearing from those. Of course, that is a big subject in itself. Then we were going to have our final recommendations to the Secretary by June of next year.
The law indicated that the Secretary is obliged to identify interim standards for e-prescribing by September of 2005. That would be then used for demonstration projects in 2006, followed by an assessment of those demonstration projects in 2006, in 2007. I think the area that people wanted it to accelerate was, the law sid that the final e-prescribing promulgation would be April of 2009. I think a lot of folks say that April 2008, we could do that earlier than April of 2008.
DR. COHN: I think we are familiar with what the MMA says. I do agree with Stan that we need to be flexible in terms of identifying low-hanging fruit and bringing it forward as seems appropriate.
I don't think that we need to answer this particular issue now.
DR. HUFF: No, we don't.
DR. COHN: But I do think that your framework was actually a very good framework, and I think we ought to have copies for the subcommittee made, just so we can make notes on it as we hear more testifiers.
I don't know that -- once again, one of the wonderful things about a work plan is that we can modify it as we go forward. I do agree with Jeff that we need to be careful not to start coming to premature conclusions, but I think there is a big difference between straw people and premature conclusions. I think we do need to begin to start framing where we think we are going. Certainly we have heard a lot of commonality so far.
DR. HUFF: Part of this is a feeling of deja vu. There are some of these issues I think we have already investigated. Are we really going to choose something different for the names of diseases than SNOMED? Indications are names of diseases, basically, diseases and syndromes. Are we really going to make a different recommendation for that? Are we going to make a different recommendation for the names of lab result names? We have already recommended NCDPD script, so some of this I would just see as a reiteration of recommendations that we have already made. People may not realize that we are mostly educating the population, who wasn't aware of what has already been done to some extent.
MR. BLAIR: The other piece is, to some extent we are giving an opportunity for people in this industry to be heard as part of the input to these recommendations, which the law required us to do.
MR. REYNOLDS: The thing I would overlay on top of what I see from Stan is, I think the Rand study today had four categories of patient safety, health outcome, helping patients manage costs, maintaining privacy, promoting clinician acceptance, and I added another one called promoting industry acceptance.
If we were to move on any kind of a faster track, if we would put that as vertical columns against that, so that if we pick NCPDP script, just to use an example, not to recommend anything, not to use their scale of one to seven and not to get complicated, but if you could check off so that the industry -- you had checkmarks in all the right places based on some of these, whereas others had questions, you start looking at a grid where we have places we have questions, we have places we have checks, to where it would at least say we have cross footed these early recommendations with what the industry -- back to your point, the industry has already accepted, clinicians may have already accepted, not going to hurt the patient, identify -- for example, patient identifier; if it is left like it is right now as a Medicare number, we would be paying everything on that.
So if you use that as a cross foot to basically say this is there, we are not reinventing the wheel, we are not doing anything that is going to act any different than it does now and it is not going to degrade any of these items, but it just gives you another way of looking at it. So as we present it to others, we aren't sitting here and hearing all the details and hearing all the issues. You have got the industry sitting around the table, to where once we put that grid up there, just like we do with letters, once we put that grid up there, if anybody challenged a check, we would talk about it.
But if that check goes all the way across those, back to what Stan put up there, you would be hard pressed not to say let's put that one down and let's move to the ones that have the questions in box six, and not flog them if they are already in place, and we have got an industry that is ready to do them, and we have got an industry that is already using them.
That is just a thought, as a nice way to present it, so that you bring everybody that is not in here today up to speed, and they can look at it and agree.
MS. TRUDEL: I just wanted to add another dimension to that, and that is something that is illustrated very well by the issue of the patient identifier.
The recommendation review, existence or lack thereof of checkmarks all the way across may be very different when you are looking from the perspective of an implementation of a drug benefit under Medicare, with the ultimate goal of trying to enhance e-prescribing in the health arena as a whole. So it may be something where an MMA recommendation needs to be made and a corollary footnote might say, however, when you start to broaden the perspective, this issue takes on a completely different perspective.
MR. REYNOLDS: Or an example I was thinking about with that, if a checkmark for MMA will also have a comment that says, if this moves into the commercial sector, whatever the patient identifier is in that sector is what is in there. So you don't have to come up with them. All insurance companies have an identifier. So those are the ways you could have it work across the board.
DR. COHN: I think we heard today about MPI activities that do the same thing. If you are 65 years of age and become a Medicare member, the HIC number history doesn't become very useful for probably a couple of years. So all of these things do work together. But we have heard a variety of possible solutions we can put up there and recognize.
Stan, how are you? You're not looking too happy at this point.
DR. HUFF: No, I'm happy. I'd dance if I could. I saw both of those comments as very constructive, friendly amendments.
DR. COHN: That's fine. I think we probably ought to try to put something together and print it out for everybody, once again, not as a decision piece, but something that allows us a framework to move forward a little better.
Other comments, other thoughts? Jeff?
MR. BLAIR: No.
DR. COHN: I think I should probably move into wrapup now. We all are finishing 19 minutes ahead of schedule, but that schedule was of course 6 p.m. Tomorrow we start at 8:30. We move into talking with software vendors most of the morning, and then on to the drug knowledge vendors in the afternoon. Then we spend a little time talking about some HIPAA issues, with once again very likely a late afternoon finish.
With that, I want to thank you all. I think it has been a very interesting day. Maria, thank you for a great job putting it together. I look forward to seeing everybody early tomorrow morning. The meeting is adjourned.
(The meeting recessed at 5:46 p.m., to reconvene the following day)